About
I'm Ritik Singla, a backend engineer in India with four years of experience. I started at BNY Mellon, building early RAG question answering on GPT-3.5. Since 2023 I've worked remotely with US teams at Lotus Management, as a founding engineer building EmailZap, an AI email assistant. Most of my work is LLM products in Python on AWS.
Work
Lotus Management
Aug 2023 to presentSoftware Engineer · Remote · formerly Studio Management LLC
An investment firm that invests in AI companies and incubates its own. I work as an engineer on companies in its portfolio, and I've mentored and onboarded 5+ engineers, reviewing their PRs and backend designs.
EmailZap: AI email assistant for Gmail
Aug 2023 to presentFounding engineer
- Rebuilt a Lambda and MongoDB prototype on Django, Celery and PostgreSQL, and took it to 10,000 signups, first paying customers in six months, and 60 to 70k emails a day.
- Moved email processing to an event-driven Gmail Pub/Sub to Lambda pipeline: new mail is handled in under a second, with no manual scaling.
- Built Zap, the in-app LangGraph agent, and the inbox search it uses (hybrid BM25 and vector retrieval with reranking). Weekly unique users rose about 3.7x in four months.
- Built the evals: a golden dataset, 15 models benchmarked on accuracy, latency and cost, and weekly drift detection. Cut LLM cost by 86% at the same accuracy.
- Moved 443 GB from MongoDB back to PostgreSQL when Mongo got too expensive at scale, and cut storage by 95%.
Thistle: meal delivery, Series B
Nov 2024 to Aug 2025Backend engineer, alongside EmailZap
- Fixed a cache stampede on the busiest public page with a single-flight rebuild lock, pre-warming and jittered TTLs.
- Built coupons, referrals and win-back systems in Django with zero-downtime migrations.
BNY Mellon
Feb 2022 to Jul 2023Pune · joined as an intern, stayed on full time
Software Engineer I
Jul 2022 to Jul 2023- Built RAG question answering over Confluence pages and PDFs for Eliza, BNY's internal AI platform, with LlamaIndex and LangChain on GPT-3.5 Turbo and open models.
- Compared fine-tuning with in-context learning for natural language to SQL, chose in-context learning, and built the chat interface in Gradio.
Software Engineer Intern
Feb 2022 to Jun 2022- Built ML models with BigQuery ML and Vertex AI AutoML on Google Cloud, tracked experiments in MLflow, and shipped the final model in a Streamlit app.
Netaji Subhas University of Technology, Delhi
2018 to 2022B.E., Electronics and Communication
College was where I tried things. I built web apps, including a college management system on Node.js and MySQL with JWT auth, did graphic design in Photoshop, and took the data structures, databases, operating systems, networking and ML courses that pulled me towards backend and AI.
Outside class, I was part of the college's IEEE student society, where I learned electronics hands-on and built hardware projects with Arduino.
I also did well at competitive programming: global rank 65 in CodeChef's August 2020 Challenge (Division 1, out of 30K), global rank 372 in Google Kickstart Round H 2020, and peak ratings of 2206 on CodeChef, 1889 on Codeforces and 2129 on LeetCode.
On the side
I build things to understand them. Right now that's Learning Concepts, my own library of computer-science courses, and Life & Lore, a journal that grew into a personal app. I'm also starting to write.
What I'm looking for
Backend and applied AI engineering roles, remote, on a team that ships LLM products to real users. If that's you, I'd like to hear from you.