About

Engineering philosophy

My projects follow a simple rule: no black boxes in the core. If Polaris does masked language modeling, the masking strategy, the objective, and the encoder are code I wrote and can defend — PyTorch supplies only tensor primitives, autograd, and optimizers. If Meridian answers a question, the tokenizer, retriever, reranker, and verifier behind that answer are all mine.

I care about the unglamorous parts of ML engineering: reproducible training runs, honest benchmarks, evaluation harnesses that can't be gamed, APIs that fail loudly, and documentation that treats the reader as a peer. A library that trains a model but can't explain its own numbers is half-finished.

Why from scratch?

Re-implementation is my learning strategy and my credibility signal. Reading a paper tells you what a mechanism claims to do; implementing it tells you what it actually does, where it breaks, and what the paper left out. Every component in my projects exists because I needed to understand it deeply enough to make engineering decisions about it.

What I'm learning now

• Faithfulness verification and domain transfer for NLI models.
• Efficient inference: KV-cache management, speculative decoding, quantization.
• Retrieval quality — where dense methods actually beat strong lexical baselines.
• Evaluation methodology: how to measure models without fooling yourself.

Where I'm headed

The grounded biomedical direction I set out toward is now real: Meridian answers questions over PubMed abstracts with per-sentence citations, entailment verification, and calibrated abstention. Next: closing its measured domain-transfer gap, scaling the corpus, and training the cited-answer generator at scale — alongside maturing Polaris and Zenith into libraries other engineers reach for when they want transparent NLP infrastructure.