AI/ML engineer and data analyst building autonomous systems that execute, then prove themselves. I work in agentic pipelines and trading bots, and I measure every result the hard way: live fills, backtests, replay, audit trails.
abhiramk2305@gmail.com / github.com/abhiram-k-2223 / linkedin.com/in/marihba
position
what was measured
result
BULLET
2,880 closed trades over a one-week soak, from a $2,000 seed
+$72,818
QUANT
1,331 parameter combinations swept, UNI/SUSHI pairs reversion
+461% / 3 y
AFRDN
6M+ PaySim transactions across the fraud-response network
99%+ ROC-AUC
MON//INDIA
15 RSS feeds + NSE corporate filings distilled into one ranked queue
Alpha 0–100
figures measured in backtest, replay, or paper trade. nothing here promises profit.
Four positions. Each one builds, runs, and reports — the figure on the right is what it measured. Three repos are public; the fourth stays private.
An autonomous memecoin trading bot for the Robinhood Chain, written in Rust. It watches the chain for new pools and handles its own entries, exits, and risk end to end, in paper trading only. The codebase stays private.
+$72,818
2,880 closed trades, one week
private repo, no public link
Pairs-trading research under a real constraint set: ₹15,000 of capital, a 30% flat crypto tax, and no directional market exposure. Only one of eleven screened pairs survived cointegration testing: UNI/SUSHI.
An autonomous fraud detection and response network. GraphSAGE embeddings, XGBoost, and a PPO reinforcement layer team up to spot and answer anomalies across the transaction graph in real time, with dashboards stakeholders can actually read.
An event-driven market intelligence terminal for Indian markets — a research assistant, not a trading bot. It never executes; it tells you what happened and why it might matter, ranked so you read the important part first.
Two roles, one degree in the books — and the numbers each produced.
mar 2025 — apr 2026
sep 2024 — sep 2025
sep 2022 — jul 2026
graduated july 2026 · CGPA 7.03
The stack I reach for across agents, models, data, and delivery.
Two hobbies, kept the way I keep everything else: on the record.
Top 10% on chess.com — 95th percentile in blitz and bullet, learned the hard way: no coach, plenty of losses, and the patience to review every one of them.
A Letterboxd habit with a five-star shelf worth defending: Ran, the Apu trilogy, The Big Short, The Darjeeling Limited, Chungking Express, and The Banshees of Inisherin.
Open to AI/ML, data, and quant engineering roles. If your team runs autonomous systems and wants them measured properly, this is the place to reach me.