Software Engineer III, AI/ML at Google

I build agentic AI systems at Google.

Founding engineer on āagman, an AI trading desk for Indian markets. Ranking and pricing models at Cars24. And a free interview playbook for engineers.

See the work Read the playbook Browse the kits

GoogleSoftware Engineer III, AI/MLOct 2026 - now
KotiLabsFounding Software Engineer, āagmanJul 2025 - Apr 2026
OpenAIFreelance contributor, OpenAI (open source)Feb 2024 - Feb 2026
Cars24Data Scientist, Ranking, Personalization & PricingDec 2023 - Jul 2025
IIT DelhiM.Tech, Computer Science and Engineering2021 - 2023

āagman · live when I left

Selected work

Products in production, and the numbers they moved.

This is selected work. See the whole career timeline →

Flagship · KotiLabs · Jul 2025 – Apr 2026

āagman, an AI trading desk for Indian markets.

Live when I left

I joined as the founding software engineer, led a team of three and owned the platform end to end, from the agents to the execution path.

01Agent workflow02Data pipeline 03Browser extension04The app

Cost per query, with an intent router in front of the full model.~~$0.10~~$0.01

How the platform works → The 8-part architecture series →

LLM calls skipped: 90%+; by a distilled Llama-3 router and a semantic cache
Screener speed-up: 10×; over sequential, across 500+ instruments
Policy layers: 5; LLM plans, OPA/Rego rules decide what runs
Brokers wired in: 4; Zerodha, Groww, Angel One and Upstox

Case study · Cars24 · Dec 2023 – Jul 2025

Personalization that replaced 25 cohorts.

Read the deep-dive →

Click Recall@50+150%

Each user got their own ranking instead of their cohort's.

Buyer conversion

+16%

under 100 ms at p99

Cold start

35%

of users had no clicks; they get a personal ranking too

Default sort · GBDT

+21%

view-to-buy-intent on the top deciles

Price model

~70%

of predictions within ±5% of the sale price

The rest of the work

Graphs, intent data and public-sector data.

Full career timeline →

Vieu · Data / ML EngineerApr - Oct 2026
Entity resolution and graph pipelines for a 900M-node person graph. Remote.
6sense · Data Scientist, intern → full-timeMay 2021 - Oct 2023
B2B intent data, NLP and computer vision. Bengaluru.
Ministry of Science & Technology · Data AnalystDec 2021 - May 2023
ETL and policy-monitoring dashboards. New Delhi.

The interview playbook

Four stages, written to help you through yours.

16 free parts from a full search: the first message, many loops at once, every round, and the choice at the end.

Stage 1: Getting the interview

Replies, referrals, and a résumé that shows the work.

Stage 2: Running processes in parallel

Several companies moving at once, so the offers land together.

Stage 3: Each round of the loop

Coding, ML domain, values, the team match and take-homes.

Stage 4: Choosing between offers

Comparing honestly, negotiating, and what the rejections taught.

The playbook is free and stays free.

No sign-up, no paywall on any part.

Start at stage 01

New series

Inside āagman: the architecture series.

An eight-part look at how the product was built, one layer per part: the strategy language, the agent desk, the backtester, the data platform, the risk gates and the execution path.

Read the series

Kits and 1:1 calls

Go further, at your own pace or with me.

Prep kits for each round, and calls when you want a second pair of eyes.

DSA round kit

250 public problems that cover the 24 patterns DSA rounds are built from, with a 118-page playbook on how to solve each pattern and what to remember, plus an 18-day sprint and an 8-week plan.

₹499

ML domain kit

A 216-page playbook with 212 question areas and answer outlines, six worked designs, a tested code pack and a 14-day plan for ML and AI domain rounds.

₹999

Agentic AI kit

200 question areas, 8 worked system designs, a 281-page playbook and a tested code pack for agentic AI and LLM system rounds.

₹1,299

Values and leadership kit

169 prompts with answer skeletons across 18 competencies, a story bank with a coverage check, a 135-page playbook and 14-day and 7-day plans for values and leadership rounds.

₹499

AI prep OS

A 25-file prompt pack, an 80-page playbook and a tracker for running interview prep with an AI assistant: coach rules, session protocols, handoffs and mocks.

₹999

Job search OS

A job search tracker, 27 message templates and an 83-page playbook for running a search while employed, from target list to signed offer, with an India-first contract checklist.

₹699

All kits

Every kit above, in one purchase.

₹2,999

1:1 calls

Mock interview60 min₹1,000 Interview prep & tips30 min₹500 1:1 mentorship30 min₹1,250 Resume review30 min₹500 Priority DM questionReply within 2 daysFree

Independent material. Not affiliated with or endorsed by any employer.

FAQ

Questions people ask.

Is the playbook really free?

Yes. Every part is free to read, with no sign-up. The playbook is free and stays free.

What is the difference between the series and the kits?

The series is the story and the method. The kits are the working material: ranked problem lists, answer outlines, plans and trackers you use day to day.

Do you take 1:1 calls?

Yes: mock interviews, interview prep, résumé reviews and mentorship. Each one is listed above with its length and price, and you book it on Topmate.

Is any of this connected to an employer?

No. The playbook, the kits and the calls are my own, and not affiliated with or endorsed by any employer.

How do I reach you?

Through the contact form. Pick hiring, collaboration or something else, and I'll reply from there.

Work with me

Building something with agents? Let's talk.

One form for everything.

Hiring Collaboration Something else