About
Building products with AI and data.
I’m Mayank, Senior Consultant, Product at MiQ Digital in Bengaluru. Seven years, one company, four titles, and a fairly consistent obsession: finding the person standing between a question and its answer, and building the thing that means they don’t have to stand there any more.
I did not start in product. I started as a business analyst in 2019, building an attribution model that measured whether people who saw a film’s digital ads actually turned up at the cinema. I spent two years being the person in the middle, the one everybody messaged when they needed a number. That is not a complaint. It is the single most useful thing that ever happened to my product instincts, because I know exactly what that work costs and exactly how invisible it is.
Everything since has been a version of removing it. An agent framework so analysts stop hunting across four systems for one answer. TESS, so a product analyst can ship a UI change without waiting three weeks for an engineer. A Healthcare AI stack so a planner gets a compliant answer, with its reasoning, instead of a meeting.
None of it needed a frontier model. All of it needed knowing the domain well enough to tell which questions are worth automating and which ones are hard for a good reason.
What I believe about building
- The queue is the cost
- Almost nobody measures the wall-clock time between wanting a change and having it. It is usually the largest number in the process and the one nobody owns.
- A refusal should be as useful as an answer
- A system that declines badly does not remove risk, it relocates it somewhere nobody is watching. In regulated domains that distinction is the whole product.
- Constraints belong in the path
- Checking output after the fact means the restricted thing was already computed, logged and cached. Harder to build in the path. Not optional.
- Enablement is the bottleneck, not tooling
- AI removed the syntax barrier to building software. It did not teach anyone that software is something you operate. That gap is taught, slowly, by a person.
Selected work
The last two years at MiQ. Where there is a number I stand behind, it is next to the work; where there is not, there is no number.
MiQ Digital · 2026
TESS: shipping UI without an engineer
An AI IDE framework that let product analysts deliver micro-frontend UI changes end to end, cutting combined frontend and backend delivery time by 85%.
Read the case study85%less delivery time
MiQ Digital · 2025–2026
A Healthcare AI stack
Rebuilding a healthcare advertising product line around AI features, where the hard constraint is not model quality, it is what you are allowed to know.
Read the case study$90M+revenue across the product line
MiQ Digital · 2024–2026
Teaching an organisation to build with AI
Early adoption of Vertex AI, Cursor and Replit inside a company that had not been built for any of them, plus the club, the onboarding and the scrum cadence that made the adoption stick.
Read the case study
The things I have not built sit separately, under Ideas.
Seven years at MiQ Digital
Senior Consultant, Product (US Local Products)
Apr 2026 – Present
- Building a Healthcare AI stack (data, compliance, retrieval and agent surfaces) as the Healthcare product line shifts to AI-native features, so measurement and audience questions resolve without analyst intervention.
- Shipped TESS, an AI IDE-based webapp creation framework: upskilled product analysts on SDLC and CI/CD and built the guardrails (typed contracts, lint/test gates, shared design tokens) that let non-engineers deliver micro-frontend UI changes such as input forms end to end, cutting combined frontend and backend delivery time by 85%.
Product Manager / Consultant (US Local Products)
Jan 2024 – Mar 2026
- Led 10+ member cross-functional teams building products that help US advertisers reach high-value Healthcare, Entertainment and DOOH audiences; delivered ~$90M in revenue including 2025 Healthcare releases (2 HCI updates, MiQ Health Segment v2) and an AI-powered Healthcare chatbot that cut analyst dependency for measurement queries.
- Built an A2A agent framework to eliminate time analysts spent navigating fragmented data systems, enabling natural-language queries across compliance, wiki and measurement via Vertex AI; adopted Replit and Cursor to cut internal tool build time; founded MiQ’s AI Club to scale AI/ML literacy across teams.
Associate Product Manager / Consultant (US Local Products)
May 2022 – Dec 2023
- Built the DOOH product suite to close advertiser gaps in out-of-home planning: aviation targeting, proximity, cross-channel, Search Sync and an automated RFP generator; grew DOOH from $2M to $5M (150% YoY, 2022–23) owning end-to-end PRD, sprint planning and stakeholder alignment.
- Replaced manual pharma client reporting with an automated activation and measurement solution (~8 hrs/week saved per client, ~100 analyst hrs per run); built an Entertainment Planning Tool so planners could discover relevant content inventory in minutes, powered by a content and collaborative filtering recommendation engine.
Senior Business Analyst
Apr 2021 – May 2022
- Replaced manual analyst workflows for $3–4M enterprise accounts with automated pipelines (PySpark, SQL) and self-serve dashboards (Google Data Studio, Apps Script), cutting recurring manual work 30–40% and freeing the team for higher-value analysis; onboarded and upskilled 5+ junior analysts.
Business Analyst (incl. Internship)
Jan 2019 – Mar 2021
- Built a Movie Online-to-Offline attribution model proving digital ad ROI for film studios by measuring actual theatre footfall; gave a cybersecurity client real-time market impact visibility via a data breach dashboard ($500K account increment during COVID); built CRM segmentation models for personalised campaign targeting across auto and CPG brands.
Before that
IIT Kharagpur
Jun 2018 – Jul 2018
Research Intern, Recommendation Systems, ISE Department
Researched graph kernel models to improve recommendation accuracy in sparse-data environments; benchmarked collaborative filtering approaches on RMSE, MAE, precision and recall. Those foundations directly shaped recommendation engine design in later roles at MiQ.
IIM Lucknow
Jun 2017 – Jul 2017
Data Science & Analytics Intern
Completed an intensive analytics programme covering R, statistics and econometrics; executed Harvard Business School case analyses and built 2 mini-projects plus a capstone on MBA salary trends, airline pricing dynamics and hotel industry performance.
R.V. College of Engineering
2015 – 2019
B.E. Industrial Engineering & Management · 8.64/10 CGPA
Relevant coursework: Product Management, Operations Management, Advanced Statistics, RDBMS, Data Analytics, Work Systems Design
What I work with
- AI & Product
- Agent frameworks (A2A)RAG & retrieval designEvals & guardrailsAI enablementProduct strategyRoadmap & OKRsPRDsA/B testingGTM
- Engineering & Data
- PythonSQLPySparkRCI/CDMicro-frontendsBigQueryFirebase
- Platforms
- Vertex AIGCPCursorReplitDV360ADHTableauGoogle Data Studio
- Ways of working
- AgileScrumStakeholder managementCross-geo delivery (US/India)
Recognition
- Finalist, Google Agentic AI Day Hackathon (2025)
- Winner, Wequity सम AI Hackathon (2025)
- MiQ Emerging Leaders Programme
- NPTEL Gold Medalist (R Programming)
Away from the desk
- Chess (Bengaluru Chess Club regular)
- Running & trekking (3 Himalayan treks)
- Speed cubing (Co-founder, RVCE Cube Society)
The cubing and the chess are probably not a coincidence. Both are pattern recognition under a clock.
Find me
I like talking to people putting AI inside organisations that were not designed for it. mayanksagar26@gmail.com is the fastest way to reach me.