Ritwik Mitra.
Building for 350M users.
Engineering leader with 20 years building consumer platforms, recommendation systems and applied AI — from 0-to-1 launches to products used by 350M+ people a month. Most recently Senior Director of Engineering for Dailyhunt & Magzter.
Products people open every day.
I build consumer products that hundreds of millions of people use, and the engineering teams behind them.
The work I'm proudest of is Dailyhunt's recommendation platform. I architected the first personalised For You feed with a three-person team, then built out the team and the platform that took it to 300M monthly users. Ranking changes we shipped raised time spent per user by 35%.
Over ~15 years at VerSe Innovation I worked across mobile VAS, carrier billing, feed & personalisation, digital publishing and data, growing from Tech Lead to Senior Director. Most recently I owned backend, notifications, data and applied AI across Dailyhunt and Magzter, leading through senior engineering leaders. I still stay close to the code through design and code reviews. It's how I mentor, and how I keep my judgement honest.
Earlier: Sun Microsystems and GlobalLogic. MSc in Computer Science, University of Delhi.
// career arc
From engineer to Senior Director.
Senior Director of Engineering · VerSe Innovation
Oct 2021 — Sep 2026- Owned Dailyhunt backend and notifications, Magzter backend and engineering tools, and applied AI across both products, with oversight of the data charter. Led through senior engineering leaders.
- Set the engineering agenda as Dailyhunt passed 350M MAU: recommendation quality, efficiency, and new revenue lines including Dailyhunt Premium and a short-stories format.
- Moved applied AI from evaluation to production for content classification, story summarisation and notification titles; made Qdrant the AI-search system of record.
- Partnered with the CFO to cut monthly infrastructure spend 49%, helped by Go services and GraalVM native builds that cut compute 30%+.
Director of Engineering · VerSe Innovation
Apr 2019 — Sep 2021- Owned the Dailyhunt feed platform, personalisation and app API gateway. Scaled the For You feed from ~80M to 300M MAU at p95 < 300 ms.
- Guided ranking strategy that raised time spent per user 35% while holding CTR and retention; grew selection & ranking strategies from 4 to 200+.
- Ran multi-armed-bandit experimentation at ~3 live feed experiments a week, and led the GCP migration of 40+ services.
Senior Engineering Manager · VerSe Innovation
Jun 2016 — Mar 2019- Architected Dailyhunt's first personalised For You feed with a three-person team, writing code and sharing product direction.
- Automated notifications on a platform processing 250M+ daily events, the foundation for billions of pushes.
Technical Delivery Manager → Senior TDM · VerSe Innovation
Jul 2013 — May 2016- Led iPayy, a carrier-billing and merchant-payments platform, from 0 to 1: product definition, architecture, team and partner integrations.
- Launched across all major Indian telecom operators; grew the team from 3 to 9.
Technical Lead · VerSe Innovation
Jan 2011 — Jun 2013- Built news and content services over SMS and USSD that reached millions of users.
Earlier · Sun Microsystems & GlobalLogic
2006 — 2010- Member of Technical Staff, Sun: enterprise management platforms (Sun Management Center, Ops Center).
- Senior Software Engineer, GlobalLogic: built the Network Analysis module of a fraud-investigation platform.
Problems worth solving.
Feeds people use every day, ranking that moves engagement, and AI shipped into real products.
Dailyhunt's recommendation platform, from scratch
Architected the first personalised For You feed with a three-person team, then built the team and platform that scaled it from ~80M to 300M monthly users at p95 under 300 ms.
Time spent per user
Ranking strategy that lifted engagement while holding CTR and retention, backed by ~3 live bandit experiments a week and 200+ ranking strategies.
Applied AI in production
Content classification, story summarisation and AI-written notification titles, live across the products. Qdrant became the AI-search system of record and replaced the legacy Elasticsearch stack.
Notifications platform
Automated notifications on a platform processing 250M+ events a day, the foundation for billions of personalised pushes.
iPayy: payments from zero
A carrier-billing and merchant-payments platform, taken from idea to launch with every major Indian telecom operator. Merchant APIs, settlements and dashboards.
Efficiency through architecture
With the CFO, cut monthly infrastructure spend from ₹45M to ₹23M (~₹260M a year) without losing scale or reliability. Go services and GraalVM native builds alone cut compute 30%+.
Dailyhunt Premium
New revenue lines, including bringing Magzter's premium magazine catalogue into Dailyhunt Premium.
What I bring to the table.
Leadership
Recommendations & product
Engineering & data
Applied AI
Off the org chart.
Unapologetic foodie
A committed non-vegetarian who will happily drive across Bengaluru for the right plate. Ask me for a recommendation — you'll get three, with opinions.
Dad, twice over
Father to a teenager and a toddler — which, it turns out, is excellent training for stakeholder management and incident response.
Still building
Weekends go to side projects: an AI debate arena where two agents argue opposite sides, and a job-matching engine with vector search and LLM reranking.
What I think about — tech, AI & leadership
// short takesLead through leaders, stay close to the code
At scale, your job is building the people who make the decisions. Design reviews are where you keep your judgement calibrated without taking the wheel.
LLMs are an infrastructure problem first
The model is the easy part. Routing, fallbacks, evaluation and per-token cost visibility decide whether AI survives contact with production.
Cost is an architecture metric
A cloud bill is a design review you didn't schedule. The biggest savings come from changing how systems are shaped, not from discounts.
Benchmarks aren't production
Every datastore looks great on paper. Test it with your access patterns, your failure modes and your on-call team before you bet on it.
Got five minutes? Take a break.
Two arcade games and two board games against an AI opponent. All built in plain JavaScript and playable on your phone.
Prefer a shell to a résumé?
Type help — or tap a command below.