Abhiram Dharme
About

Hi, I’m Abhiram.

Abhiram Dharme

IIT Delhi · 2026

I’m a dual-degree (B.Tech + M.Tech) student in Computer Science at IIT Delhi, graduating in 2027. I spend most of my time on deep learning architecture and training techniques where model design and engineering details both matter. I am also interested in scaling laws and data scarcity.

Lately, most of my time has gone into CycloFormer, a rotation-invariant transformer for hand-pose estimation from wrist surface electromyography. It is joint work with Prof. Chetan Arora at IIT Delhi and Ankush Gupta at Google DeepMind, and is currently under review at NeurIPS 2026.

I’m drawn to work that combines careful experimentation, strong engineering, and a taste for hard technical problems. Recently, that has meant wearable biosignals, equivariant models, scaling behaviour, and ML systems.

Recognition
JEE Advanced 2022
top ~0.5% of 250,000 candidates
AIR 1055
JEE Main 2022
top ~0.14% of 900,000 candidates
AIR 1342
GATE CS 2026
AIR 929
Codeforces
top ~10% globally
Expert · 1672 peak
Tower Research Capital — Data Challenge
institute-wide · 2024, 2025
Top 50
Inter-IIT Tech Meet
IIT Delhi contingent · 2024, 2025
Tech 13 · 14
Experience
May 2026 — Jul 2026

Google

Software Engineering Intern · Bangalore

Worked on Gemini Cloud Assist and Google Cloud AI Diagnostics and Monitoring.

  • Designed and implemented an MCP endpoint enabling Gemini Cloud Assist to auto-diagnose and resubmit failed enterprise batch jobs (e.g., OOM failures), replacing manual log-debugging with one-click resolution — a step toward agentic, self-healing cloud operations.
  • Migrated a monitoring library from Cloud Monarch to Google’s company-wide GMon framework, reducing internal tech debt and improving production-metric observability across legacy systems.
May 2025 — Jul 2025

Coinbase

ML Engineering Intern · CBGPT · Bangalore

Built a knowledge-gap classification pipeline for the CBGPT customer-support assistant — LLM-based topic modeling with a FAISS retrieval layer and batched entailment, surfacing under-documented support areas at scale.

Jan 2024 — Mar 2024

Torch Investments

ML Engineering Intern · New Delhi

LightGBM and CatBoost models for Fortune-500 equity return prediction inside a live $3M+ strategy; 2014–2024 backtest of the pipeline yielded 21% CAGR.

Leadership

Inter-IIT Tech · Arista Networks

Project lead · IIT Delhi contingent

I led IIT Delhi’s effort on Arista’s dense WiFi resource-management problem statement. My contribution was project leadership rather than claiming ownership of every modeling component: keeping simulation, modeling, evaluation, and the final technical narrative pointed at the same operational problem.

Off the keyboard
  • Cricket — playing on weekends, watching whenever ODI/Test happens to be on. Not a big fan of T20 though.
  • Counter-Strike 2 in the evenings, with friends — the best ROI for time spent we ever found.
  • Singing — Indian classical music mainly • Shankar Ehsaan Loy • Bhimsen Joshi
  • Anime - I watch anime in bursts, either seasons at a time or no episodes at all.
Abhiram at the archery range, IIT Delhi.