01 / 11 — ABOUT — WHO WE ARE
Gigaflop Techlab · A DiscoverWebTech company · Est. 2012
Rev. 2026.05 · v3.0

Built on 14 years
of engineering.
Refocused for data + AI.

Two co-founders. A 14-year engineering bench. One accountable team from kickoff to ship. DiscoverWebTech (est. 2012) is our parent firm still operating, still shipping. We launched Gigaflop Techlab in 2019 as the dedicated data + AI practice for SaaS and D2C teams: same founders, same bench, productized D1–D5 / A1–A5 engagements. Audit-first, milestone-paced, no offshore handoffs the methodology has been constant since 2012.

Book a 30-min discovery call → Read our work →
Established 2012 200+ systems shipped 90% retention on retainer US · UK · EU · APAC
001 / Engineering depth
14 yrs
Through DiscoverWebTech, since 2012
002 / Systems shipped
200+
Production data and AI engagements, cumulative
003 / Retainer retention
90%
Year-over-year on D5 / A5 accounts
004 / Founder-led
100%
The architects on the call are in the working sessions weekly
Section03
— Story · DWT to Gigaflop

What we built
and why we launched a sister firm.

14 years, two practices ↘

DiscoverWebTech was established in India in 2012 by two engineers who'd been frustrated with how international clients were served by Indian agencies. The default model was opaque buyers met an account manager in their timezone, the work was done by people they'd never speak to, the architects on the proposal weren't on the implementation. The default firm scaled by hiring junior bodies and inserting them into deal flow.

We built a different shape. Practitioners-led - the people who scope the work do the work. Founder-on-every-call - for the first 8 years, every engagement had a founder in working sessions weekly. Senior bench by default - average engineering tenure on the team is 9 years and we don't bait-and-switch.

DiscoverWebTech's first several years were web development, custom software, and increasingly: data engineering. By 2018, data work was more than half of active engagements distinct enough from the broader software practice that the brand was carrying two different sales motions, two different client mental models, and two different positioning needs.

Rather than rebrand DiscoverWebTech, in 2019 we launched Gigaflop Techlab as a dedicated sister firm same founders, same engineering bench, focused initially on data engineering for Series A–C SaaS. We added the AI engineering practice in 2023 as production AI (post-GPT-4, post-Claude) became viable for serious customer-facing builds. DiscoverWebTech still operates as the broader software-engineering firm: web, mobile, custom builds, platform work, and longer-running engagements that don't fit the Gigaflop ICP.

Two firms. Shared spine. Both share the same founders (Siddharth and Shiwali), the same 14-year engineering bench, and the same audit-first methodology. The split is by ICP and engagement shape not by team. When a Gigaflop engagement needs full-stack product engineering (a common pattern in A4 builds), the DiscoverWebTech bench is staffed onto the same project. You're talking to one team, even when the brand surface differs.

One accountable team. Two brand surfaces. Same methodology, since 2012.

— The two firms · side by side Same founders · same bench · two ICPs
— Parent · Est. 2012

DiscoverWebTech

The 14-year full-stack engineering firm. Where the bench was built.

Founded
2012 · 14 years continuous
Focus
Full-stack software engineering web, mobile, custom builds, platform work
ICP
Mid-market software teams broadly · longer-running product engagements
Engagements
Project-based · scoped per-client · flexible structures
Status
Operating
Site
discoverwebtech.com →
— Specialist practice · Est. 2019

Gigaflop Techlab

The dedicated data + AI practice. Productized, audit-first, SaaS/D2C-focused.

Founded
2019 · 7 years as dedicated practice · AI added 2023
Focus
Data engineering + AI engineering · two practices, one team
ICP
Series A–C SaaS · D2C / e-commerce · $5M–$50M ARR · audit-first
Engagements
Productized: D1–D5 (data) & A1–A5 (AI) · fixed-scope, fixed-price
Status
Operating · accepting Q2
Site
gigafloptechlab.com →
Shared between both firms
Founders
Siddharth Mishra · Shiwali Ratan Mishra - same on both, code-shipping practitioners
Engineering bench
~24 senior engineers · 9-yr average tenure - same DWT bench staffed across both firms
Methodology
Audit-first · founder-led · milestone-paced · production-grade · no offshore handoffs · fixed-price
Headquarters
India · USD-billed · serving US · UK · EU · APAC since 2012
Cross-staffing
When a Gigaflop A4 build needs full-stack product engineering, the DWT bench is on the same project one team, one accountability
— Why "Gigaflop"
A gigaflop is a unit of computational throughput one billion floating-point operations per second. It signals what the practice does: production-grade engineering measured in real units. It's also a quietly self-aware name. A gigaflop sounds large but is, by 2026 standards, modest. We picked it because we'd rather under-promise the brand and over-deliver the work.
Section04
— Philosophy · why one team

Most agencies pick one.
We do both because
stitching is where projects die.

~22% of builds
span both ↘

The most common shape we see in the market: a company hires a "data agency" to fix the warehouse, and an "AI agency" to ship the chatbot. Six months later, the AI feature can't see the data it needs because the two firms designed for different abstractions. Three weeks of integration becomes three months. The board starts asking why.

We exist to remove that seam. The audit (D1 or A1) considers both practices. The build engagement spans whichever practice the work demands. The retainer covers both. The same engineer who designed the dbt model can call the agent that uses it.

— Practice 01

Data Engineering

Pipelines, warehouses, dashboards, and cost optimization for Series A–C SaaS teams.

Tiers
D1 → D2 → D3 → D4 → D5
Range
$4.5K–$50K · $5K–$15K MRR
ICP
B2B SaaS · Series A–C · $5M–$50M ARR
Lead
Shiwali Ratan Mishra
Data practice →
— Practice 02

AI Engineering

Chatbots, agents, copilots — plus the eval pipelines and red-teaming that keep them running.

Tiers
A1 → A2 → A3 → A4 → A5
Range
$5K–$120K · $2K–$10K MRR
ICP
D2C / e-commerce + B2B SaaS · $5M–$50M ARR
Lead
Siddharth Mishra
AI practice →
— Combined engagements ~16% of audits are combined D1 + A1. ~22% of build engagements span both practices (e.g., a D2 pipeline that feeds an A3 agent). Same proposal, same project plan, single accountable team — not two vendors stitched.
Section05
— Founders · who leads

The architects on the discovery call
are the people in the working sessions.

Practitioners first ·
Both write code weekly ↘

Both co-founders are practitioners first. Both write code weekly. Both are in client Slacks daily during active engagements. The first time you meet them is the discovery call; you'll meet them again every week the engagement is live.

SM
Co-founder · AI Engineering · 14 years

Siddharth Mishra

"Production AI without the science project."

Siddharth co-founded DiscoverWebTech in 2012 after several years building production systems for international clients. For the firm's first decade, he led architecture across web, mobile, and platform engagements work that taught him how to ship software that runs after the agency leaves.

Since 2023, his focus has been AI engineering chatbots, agents, copilots, and the eval/security/operations layer that production AI requires. He's led 30+ AI agents to production over the last 24 months, including the engagements that led to our published case studies. He red-teams AI systems for security flaws as a default part of every audit. He cares more about an eval pipeline than a pretty demo and considers most "agents" in 2026 to be elaborate prompts pretending to be more.

He's the person who'll tell you, on the discovery call, when an in-house hire is the better path than engaging us. About 1 in 5 conversations end this way.

— Specialties

  • AI agents (multi-step + tool use, production)
  • Production AI security and adversarial red-team
  • Eval pipelines and drift detection
  • Provider-agnostic architecture
  • AI product engineering (full-stack feature build)
→ mishrasiddharth.com → linkedin
SR
Co-founder · Data Engineering & Strategy · 14 years

Shiwali Ratan Mishra

"The unsexy plumbing that makes the rest work."

Shiwali co-founded DiscoverWebTech in 2012 alongside Siddharth. For most of the firm's history, she's led the practices that don't get the splashy case studies data infrastructure, BI implementation, the long-term retainer relationships that quietly compound into the firm's strongest references.

Her practice now focuses on data engineering for Series A–C SaaS: warehouse optimization, dbt and pipeline architecture, semantic layers, and cost discipline. She's the lead architect on the firm's 14-year data engineering bench and runs the strategy practice for cross-cutting client engagements (when an audit recommends a hire instead of a build, she's the one who writes the JD).

She's the person who'll tell you, when you're sure your data stack is fine, exactly which silent join is breaking your CFO's trust. About 30% of D1 audits surface findings the client team didn't suspect existed.

— Specialties

  • Snowflake / BigQuery / Redshift cost optimization
  • Production data pipelines (dbt, Airflow, Dagster)
  • Semantic layer design (dbt SL / LookML / Cube)
  • BI implementation (Looker, Metabase, Hex)
  • Strategic engagement design and proposal architecture
→ shiwaliratanmishra.com → linkedin
Section06
— Team · DiscoverWebTech engineering

The team behind
the team.

~24 senior engineers ·
9-yr avg tenure ↘

Gigaflop's two co-founders are the surface. The depth is the DiscoverWebTech engineering bench the senior engineers who staff every Gigaflop engagement alongside the founders. The bench has been growing for 14 years; average tenure is 9 years.

— Engineering bench
~24
Senior engineers across data + AI + full-stack
— Average tenure
9 yrs
Across the bench (DWT has 14-year history)
— Staffing model
2 + 1–3
Founder-led, bench-supported per engagement
— External / contractor
0%
We don't subcontract. The team that scopes is the team that ships.

When we say "no offshore handoffs," we mean: the team that scopes is the team that ships. We don't hand work to a separate delivery team in a different city or a different time zone or a different pay grade. The architects on your call own your engagement; the bench is staffed under their direct technical leadership.

— Where we are DiscoverWebTech is headquartered in India. The engineering bench is fully in-house, on-payroll, and bench-billed in USD. We've been working with international clients (US, UK, EU, AU, SG) since 2012 the operational rhythm is mature, not improvised. Standard practice: 4 hours of overlap with your timezone, committed in writing in the proposal.
Section07
— Values · stated as practices

Our values, stated as
concrete practices.

5 commitments ·
each one testable ↘

Most companies have values. Most companies' values don't show up in any specific commitment. Ours do. Each one below is operational you'd notice if we stopped doing it.

— 01

Audit-first

The practiceEvery engagement starts with an audit. We don't price builds without a 2-week diagnostic. Every audit ends with a recommendation that may or may not include a build.
You'd notice if we stoppedif we ever quoted a fixed-price build off a discovery call alone.
— 02

Honest about not-fit

The practice~30% of audits end with us recommending against the build, smaller fix, or in-house hire. We've turned down ~$400K of build work in the last 18 months.
You'd notice if we stoppedif every audit we delivered concluded "you should hire us for the build."
— 03

Founder on every call

The practiceA co-founder is in every client engagement weekly during active build phases. Not by exception by default.
You'd notice if we stoppedif the names on the discovery call weren't the names on the working session.
— 04

Production-grade by default

The practiceMonitoring, eval, runbooks, and post-handoff documentation ship with every engagement. We don't leave prototypes in production.
You'd notice if we stoppedif your team couldn't run our system after we left.
— 05

Fixed-price, no overruns billed

The practiceEvery engagement is fixed-price, fixed-scope. Overruns happen (~1 in 14) and they're our problem. We don't ask for change orders to cover internal mis-scoping.
You'd notice if we stoppedif you ever got a "scope expansion" email in week 6.
Most agencies' values are aspirational marketing. Ours are operational commitments. The difference is whether the team would push back if we tried to abandon them.
— From the firm charter
Section08
— Coverage · where we work

Headquartered in India.
Working with teams across
US · UK · EU · APAC.

14 yrs of muscle memory
on async-first work ↘

Time zones are an operational design problem, not a fact of life. We've been working with international clients since 2012 and built our delivery rhythm around overlap windows, async-first communication, and weekly synchronous demos.

Americas
US · Canada · LatAm

Overlap
4 hours, typically 6:30 PM IST onwards (matching US morning)
Demos
Friday 9 AM PT / 12 PM ET / 6:30 PM IST

UK · Europe
Middle East

Overlap
6 hours, comfortable working window
Demos
Friday 1 PM GMT / 6:30 PM IST

Australia · NZ
SE Asia

Overlap
4 - 6 hours
Demos
Friday - matching client preference

India
South Asia

Overlap
full
Demos
Standard business hours
— Operational language
English (default for all engagements, written + spoken)
— Read-only support
German · French · Spanish · Hindi · Portuguese (for client documentation review where needed)
— Operational maturity, not improvised We have 14 years of muscle memory on async-first work, written-default communication, weekly demo rhythm, and stakeholder management across timezones. Time-zone friction shows up in the first 2 weeks of an engagement and disappears thereafter. Ask any of our clients in the references section.
Section09
— Trust · recognition & posture

The boring
trust signals.

For procurement
diligence ↘

Buyers in procurement asked us to put these in one place. Below are the operational and security commitments that matter for enterprise diligence.

— Credentials & insurance

  • Established2012 (DiscoverWebTech parent firm)
  • Legal entityDiscoverWebTech Pvt. Ltd. (registered India)
  • Professional liability insurance$2M aggregate · $1M per claim
  • General liability$1M per occurrence
  • Cyber liability$1M aggregate (in process; current placeholder $500K)

— Compliance posture

  • NDAs · MSAs · DPAsstandard signing; we have templates reviewed by client counsel; we sign on your paper as well
  • SOC 2 Type 2in process · target Q3 2026
  • HIPAA BAAsavailable · current engagements include 2 HIPAA-covered builds
  • GDPR · data residencyEU and UK deployments handled via regional cloud regions
  • Data retentionclient data on our systems retained per engagement DPA, default 90 days post-handoff
  • Subprocessorsdisclosed in DPA; current list available on request

— Memberships & references

  • Industry membership. Active member of relevant data + AI engineering communities (specifics on request, with active partnerships rotating).
  • Public references. Available post-discovery-call, with prior client consent. We don't list logos publicly without explicit permission ask and we'll route the relevant introduction.
  • Client references. See /work for case studies; direct client conversations brokered after qualification.
Section10
— Hiring · when we're looking

When we hire,
we hire deeply.

2–4 senior engineers
per year, on average ↘

We're not always actively hiring. The bench is built deliberately we'd rather grow at 10–15% per year with senior people than at 40% with juniors who staff bench-time. We add 2–4 senior engineers per year on average; we've added 0 in some years.

— What we look for

  • 7+ years engineering experience, with at least 4 in production data or AI systems
  • Comfort with the full path: scoping, building, shipping, operating, handing off
  • Written communication strength we work async-first; client trust is built in writing
  • Track record on engagements where the engineer's name was on the proposal

— What we don't optimize for

  • Specific framework or vendor expertise we're stack-agnostic; we hire engineers who learn fast
  • Size of past employer strong engineers come from 5-person and 5,000-person firms alike
  • Pure CS theory background without production scars

— How to apply

If you're an engineer who's read this far and recognizes yourself in the above, email hello@gigafloptechlab.com with subject "engineering bench" and a one-page resume. We respond to all applications within 4 business days. We don't have a public careers page because the volume isn't worth maintaining one.

Section11
— Start · talk to us
/about · END ↘

We'd rather show you
than tell you.

Most companies' "about" pages overpromise. Ours describes what we do. The fastest way to find out if it matches reality is a 30-minute discovery call. No slides.

Book a 30-min discovery call → Read our work → hello@gigafloptechlab.com
P.S. 14 years of engineering means we've already had the embarrassing failures. The lessons are baked in. The case studies just don't talk about them that's the only marketing concession we make.
Section12
FAQs · common questions

Questions we get
asked most often.

5 questions ·
honest answers ↘

Yes, effectively the same team different brand surfaces for different buyer contexts. DiscoverWebTech (est. 2012) is the parent firm: full-stack software engineering, web, mobile, custom builds. Gigaflop Techlab (est. 2019) is the dedicated data + AI practice under the same founders and the same ~24-person engineering bench. If you're a Series A–C SaaS or D2C team with a data or AI problem, you're in Gigaflop territory. If you need broader product engineering, you're in DiscoverWebTech territory. In practice, about 22% of Gigaflop builds pull in the DWT bench for full-stack product work one project plan, one accountability layer.

It's a commitment, not copy and we'd notice if we violated it. A co-founder is in every client engagement weekly during active build phases. That's Siddharth on AI engineering engagements (A1–A5) and Shiwali on data engineering engagements (D1–D5). Both write code weekly. Both are in client Slacks during active phases. The person who scoped your engagement is the person in your Monday working session. We've maintained this since 2012 and specifically listed it as a testable value: you'd notice if we stopped because the names on the discovery call wouldn't match the names on the calendar invite.

Yes — and here's why that protects you. Every engagement starts with a D1 (data audit) or A1 (AI audit): a fixed-scope, fixed-price 2-week diagnostic. We don't quote build work without it. The reason is practical: about 30% of audits end with us recommending against a full build a smaller fix, an internal hire, or a different sequencing. If we quoted builds off discovery calls, we'd be pricing against assumptions. The audit is also how we catch the things you didn't know were broken. About 30% of D1 audits surface findings the client team didn't suspect. You're paying for certainty, not gatekeeping.

14 years of muscle memory, not improvisation. We commit 4 hours of overlap with your timezone in writing in the proposal. For US clients (~55% of our active engagements) that's typically 6:30 PM IST onward, matching US mornings. For UK/EU clients (~25%), we have a comfortable 6-hour working window. Written-default communication, weekly synchronous demos, and async-first delivery rhythm have been standard practice since 2012. Time-zone friction is real in week 1 and essentially gone by week 3 ask our existing clients, whose introductions we broker post-qualification.

Talk to us we'll tell you honestly in 30 minutes. Our stated ICP is Series A–C SaaS or D2C / e-commerce between $5M–$50M ARR with a data or AI problem. But ICP is a targeting heuristic, not a hard door. We've worked with pre-Series A teams with strong technical clarity, with enterprise teams who needed a focused specialist practice rather than a consultancy, and with companies outside those revenue bands when the engagement shape was right. About 1 in 5 discovery calls ends with us pointing you elsewhere we'd rather do that in 30 minutes than let a mismatch drag through proposals. Book the call and find out.