Access to talent - surge capacity & specialized skills
IIT Madras
Core delivery team - batchmates
10+
Companies we've added AI leverage to
Portfolio
One team for the whole stack. We build websites and mobile apps, AI apps and agents, the models behind them, and the hardware and cameras they run on. Here is some of it in production.
BuiltOptimisation models · Forecasting · Data platform · Dashboards
Informed by David's work at Target: moving goods from Asian origin ports to nearly 2,000 US stores at the lowest landed cost without missing the shelf date.
Port and lane selection, trading transit time against cost and capacity
West Coast direct versus the Panama Canal to Gulf and East Coast ports
Demand-aware allocation from distribution centres to stores
Forecasts that flag delays while there is still time to reroute
BuiltMobile app · On-device AI models · AI coach · Web platform · Trainer app
An AI personal trainer on the phone. The camera recognises the exercise, counts reps and scores every one of them in real time, with no wearables. Then it explains what to fix and builds the plan around you.
Live exercise recognition, rep counting and per-rep form scores
Post-session AI analysis: movement-quality checks and best and worst key frames
An AI coach that knows your sessions, scores and goals
Assessments, programs, progress analytics, games, and a trainer app for coaches
More of our work is under NDA. Ask on the discovery call and we will walk you through what we can show.
What we do
Two bundles of work - Build for teams that need custom AI engineering, Apply for teams that need AI turned into business outcomes. Most engagements pull from both.
Build - Custom AI engineering
AI Agents
Stateful, tool-using agents wired into your stack - internal copilots, customer-support automation, sales-ops bots, research workflows. Built on modern agent frameworks with eval harnesses, not prompt-glue.
AI Foundation Models
Fine-tuning, RAG stacks, evaluation harnesses, domain adaptation. Build-vs-buy advisory before you commit to a closed API or open-weight stack. Cost and latency engineered alongside accuracy.
AI Audit
A full read of your current AI stack - what you have, what works, what doesn't, what's missing. Latency / cost / accuracy benchmarks, vendor and model evaluation, build-vs-buy analysis, security and compliance review. Output: a written report with prioritized recommendations and a 6 / 12-month AI roadmap.
AI Team Stack
Plug an AI team member into your org - AI consultant, researcher, HR / talent coordinator, ops manager, customer success. Pick the role, we deliver the person, backed by our 70k+ talent network. Faster than hiring, lower-friction than a long retainer, ramp up or down on demand.
MLOps, Pipelines & Evaluation
End-to-end ML engineering. Data ingestion, training infra, inference services, CI/CD for models, plus eval harnesses, A/B testing, drift detection, monitoring, and the dashboards that tie it all together. Production-grade and observable - not notebooks held together by hope.
AI Labeller
Domain-specific data annotation at scale - bounding boxes, polygons, video events, NLP, audio. Active learning to prioritize labels that move the needle, semi-supervised pseudo-labeling to multiply effective dataset size, inter-annotator quality tracking. Backed by the 70k talent network for surge capacity.
Computer Vision Usecases
Production-shipped CV across real-time tracking, multi-camera setups, retail and security analytics, sports analytics, industrial inspection, manufacturing QC, agritech. Calibration, homography, 3D reconstruction - and beats published academic baselines on in-domain tasks.
Hardware AI Usecases
Edge inference for real-world devices: smart cameras, IoT sensors, robotics, embedded automation, on-device assistants, line-calling and broadcast hardware. Quantization, distillation, on-device inference for ARM, Coral TPU, Jetson, RPi CM5. Real latency budgets, real power budgets.
AI Native Apps
AI-native applications - web, mobile, internal tools, customer-facing surfaces - with AI baked into the UX, not bolted on. Figma → live in days, AI-augmented build pipelines, conversion-tuned. This site is the proof.
Apply - AI for business outcomes
AI Marketer
Content production engines, customer-voice synthesis, persona research, campaign automation, performance feedback loops. The AI marketing stack, integrated into your existing tooling - not bolted on as a side project.
AI Customer Support Bot
Conversational support over your docs, ticketing system, and product knowledge. Tier-1 deflection, smart escalation to humans, learns from every resolved ticket. Integrates with Zendesk, Freshdesk, Intercom, WhatsApp.
AI Caller
Voice AI for outbound and inbound calls. Lead qualification, appointment booking, customer surveys, support triage, collections reminders. Real-time conversation tuned for Indian accents and languages, integrated with your CRM and dialer.
AI Sales Agent
Outbound SDR automation - prospect research, personalized outreach, follow-up sequences, meeting booking. Pipeline intelligence - lead scoring, deal-risk flagging, account-based research synthesized for your reps.
AI Customer Analysis Engine
Segmentation, churn prediction, LTV modeling, behavioral insight. Plugs into your CRM and warehouse. Output: actions for your sales, marketing, and product teams - not yet-another-dashboard.
AI Hiring & HR
Résumé shortlisting with reasoning, candidate scoring, interview prep, AI screening calls. Built to be transparent and auditable - no black-box rejections. Cuts hiring-team load by 70%+ on typical funnels.
AI Personal Assistant
Executive copilot for founders, GMs, and senior operators. Calendar, email triage, research, meeting prep, financial summary, learning, fitness. The high-leverage personal AI stack, set up and tuned to one operator's workflow.
AI R&D
POCs, evaluations, build-vs-buy advisory, technology scouting. Cheaper than a research hire, faster than figuring it out yourselves. Useful before you commit headcount or platform spend.
AI Services (applied)
Workflow automation, RAG over your docs, custom GPTs, internal copilots, scrape + LLM pipelines. The applied bread-and-butter - high leverage, low ceremony, fast to ship.
Why work with us
01
8 years of production AI
Computer vision and ML systems shipped on real hardware in real environments. Real-time inference on edge devices, multi-camera pipelines, model compression. Not slideware, not LinkedIn theory.
02
IIT Madras-led core team
The delivery team is built from IIT Madras batchmates with active experience across CV, NLP, full-stack, and infrastructure. Not stitched-together freelancers - engineers who've shipped together.
03
Access to 70,000+ talent
Mad About Sports alumni - a 70k+ bench of ML, data, analytics, and content talent we tap for specialized skills and surge capacity. We can scale a project from one engineer to a small pod when the work demands it.
04
AI leverage at 10+ companies
Sports-tech, hardware, analytics, edge inference - across cofounder, contractor, and advisor roles. Track record across the full stack: research, engineering, deployment, and post-launch operations.
The team
The core delivery pod. IIT Madras-led, with a 70,000+ talent network behind us for surge capacity and specialized roles.
David Gladson
Founding Partner, Chief AI Scientist
Ayudh Sharma
Founding Partner, Senior Project Manager
Vikas Raju
Senior AI & Hardware Researcher
Nitin Kumar
Senior Fullstack Developer
Who we work with
Founders building AI-first products
You need AI capability without hiring a team. We come in as a full-stack AI partner - strategy, build, deploy - until your in-house team is ready to take over.
Companies adding AI to existing products
You have a real business and need an intelligence layer - agents, customer analysis, internal copilots, automated workflows. We integrate, we don't replace.
VCs and accelerators
You back portfolio companies that need execution partners on AI. We work with multiple portcos, share patterns across them, and give honest build-vs-buy assessments.
Hardware-first teams
You need real-time inference on Jetson, Coral, Pi 5, or custom silicon. Frames per second per dollar matters. We've shipped this exact problem before.
Companies with stuck POCs
"It works on a server but we can't ship it to the device" or "the model works but the product doesn't." We unstick.
How we engage
01
Discovery call - 30 min, free
Tell us what you're building, what's stuck, and what success looks like. We tell you honestly whether we're the right team for it - and if we're not, we'll point you somewhere better.
02
Written scope & fixed quote - 3 working days
Milestones, deliverables, fixed price. No hourly billing on productized work. You see the entire engagement on one page before you commit.
03
Milestone delivery
Slack/WhatsApp updates, weekly demo, transparent progress against the plan. You can stop at any milestone - no lock-in.
04
Handover
Code, models, benchmark suite, deployment scripts, written handover doc. 2-week post-handover support included on every package.
What's the hard problem?
The catalog above is what we typically ship. The more useful conversations usually start the other way around - with the problem your team is actually stuck on. Some of the ones we love hearing about:
Restaurants / F&B
"Our CC cameras capture every table, every minute - but we still can't tell which tables are camping for hours, where servers are slow, or where in the customer journey we're losing 20 minutes."
Computer-vision pipeline that tracks table state, dwell time and service flow - surfaces the actual bottlenecks instead of slide-deck guesses, runs on the existing camera infrastructure.
D2C / Performance marketing
"We're spending ₹2+ crores a month on Meta Ads. Our 15-person performance team can't audit every campaign - bad ads stay live for days, the team is buried in dashboards."
Ad-management agent that watches creatives and performance in real time, kills underperformers automatically, surfaces what to scale. Peer benchmark: 15-person teams running at 20× volume with 5 humans.
Healthcare / clinics
"Our front desk takes 200+ calls a day. Receptionists can't triage emergency vs refill vs complaint fast enough. Urgent patients wait. The team is burning out."
Voice agent that triages, books, and escalates - with patient context memory, so it actually knows whose call this is. Indian-accent fluency built in, integrated with the existing booking system.
Lending / collections
"We need to reach 5,000 borrowers a month. Human agents burn out, scripts sound robotic, repayment rates are dropping. The voice vendors we tried sound foreign - borrowers just hang up."
Indian-language voice agent tuned for emotional nuance - scriptable but adaptive. Same stack as our AI Caller demo, deployable in 4-6 weeks.
Retail / security
"Hundreds of CCTVs across our stores. We find out about theft three days later by counting inventory. Loss is in seven figures and growing."
Real-time CV anomaly detection on the existing camera feeds. Flags happen in seconds, not days. Runs on-prem so no footage leaves the building.
Construction / supply chain
"We sign for 90-day delivery and routinely run 30 days late. Procurement, labour planning, supply chain - none of it talks to a forecast. Every delay is a surprise."
Predictive analytics on contract → delivery cycles. Forecasts feed procurement triggers and supervisor scheduling - risk surfaces weeks before it becomes a missed deadline.
Operations / leadership
"I'm in 8 meetings a day, 200 Slack threads, 50 WhatsApp groups. Critical decisions get lost in the noise. I'm always one step behind."
Personal AI that summarises every channel, surfaces the three things that actually need attention, drafts replies in your voice. Not another inbox to manage - fewer.
Off-catalog
"We have something nobody else is offering us - but we're convinced AI can solve it. We don't even know who to ask."
Our favourite kind. Bring it to the discovery call, even if it's nowhere on the catalog. Most of our best engagements start here.
