Pramod R Bhat
ML engineering for financial markets — currently building a production algorithmic trading system on NIFTY 50 stocks. Forecasting forward returns with gradient-boosted ensembles, walk-forward CV, and conformal prediction.
Most of the trading system was written between 11pm and 2am.
Snapshot
Three projects. One that matters.
- 032022 — present
enterprise process automation
Production UiPath workflows at enterprise scale. Shipping automation that organisations actually depend on — not the side of the resume that predicts the next role, but the reason the work compounds.
· UiPath· orchestration· .NET
Building things and breaking them honestly.
I build ML systems for financial markets. Day work is enterprise automation; the trading system is on weekends and late weeknights. The work I keep coming back to is the numbers that turned out to be wrong.
Most of what I've learned came from building things and then breaking them honestly. My V3.1 trading model reported 54.3% directional accuracy — well above what a coin flip would give you. After catching two leakage bugs — places where the model was effectively peeking at the answer — V5's honest baseline was 49.6%. V7 is the systematic recovery. The version history is documented and the bugs are named — I think the willingness to do that is the work.