experience

drive powerline

ml eng intern

used ml to analyze the energy market

details

built foundation time-series model for cross-region aemo predictive signaling & voltalility forecasting across diverse factors (demand, solar pv, price elasticity, price) to inform optimal energy trading. improved price spike forecasting from f1 0.40 -> 0.61 to predict critical >$1K events. built internal optimization software suite that reduced ai dev cycle from >10 days to <1 day.

berkeley ai research

ml researcher

web agents · inr uncertainty (missing-wedge)

details

built agents to navigate/analyze sites (reddit, gitlab, e-com) with eval harnesses; ongoing work on epistemic uq for inrs in tomography.

machine learning @ berkeley

consultant

cs198-126 lecturer · guardian image search

details

lectured ~40 students as lead student facilitator for berkeley's deep learning for computer vision course (cs198-126). pm/lead engineer for development of image search engine for the guardian (openclip + qwen2.5).

hhmi janelia (turaga lab)

research intern

ml for microscopy and holography · chromatix dev

details

developed ml-powered 3d snapshot microscopy algorithm w/ depth-adaptable phase mask for zebrafish brain imaging; demos + chromatix contribs; transformer-based computer generated holography for optogenetics (boosted accuracy from 60% to 77%).

vytal.ai

founding ml engineer

eye-tracking < 2° error · pytorch + aws

details

built an eye-tracking pipeline on aws ec2 for neuro-diagnostic biomarkers; optimized for reliability and low-latency inference.