Work
Projects
Open-source machine learning infrastructure, built from first principles and shipped to PyPI. Each project page reads like documentation: problem, architecture, design decisions, measured benchmarks, and roadmap.
01Grounded RAG
Stable · v1.0Meridian
A from-scratch grounded RAG engine over biomedical literature (PubMed). Custom BPE tokenizer, dense retriever, HNSW/IVF ANN, cross-encoder reranker, and NLI faithfulness verifier — no third-party pre-trained models. Every answer is cited, verified, or refused.
System pipeline
- BM25
- Dense retrieval
- HNSW / IVF
- Cross-encoder
- Cited generation
- NLI verify
- Abstain
Measured results
- R@5 after reranker rank fusion
- 0.983
- HNSW recall@10 at 0.26 ms
- 0.996
- NLI verifier, SNLI dev
- 78.3%
- R@5 after reranker rank fusion
- 0.983R@5 after reranker rank fusion
- HNSW recall@10 at 0.26 ms
- 0.996HNSW recall@10 at 0.26 ms
- NLI verifier, SNLI dev
- 78.3%NLI verifier, SNLI dev
pip install meridian-rag02Encoder NLP
Stable · v1.1Polaris
A production-inspired NLP engineering platform for understanding text. The complete encoder-side stack — data, tokenization, collation, models, training, evaluation, deployment — from scratch on PyTorch tensors.
- IMDB ceiling, four-lever study
- ~86%
- epoch-1 val with MLM (vs 0.74)
- 0.81
- versioned PyPI releases
- 17
03Decoder LLMs
Stable · v1.0Zenith
A from-scratch generative NLP library — decoder-only language models and text generation. Llama-style internals (RoPE, RMSNorm, SwiGLU), KV-cached generation, speculative decoding, LoRA, instruction tuning. Matches the nanoGPT baseline.
- bits/char, nanoGPT parity
- 2.08
- fewer target passes, spec. decoding
- 3.7×
- smaller weights with int8
- 4×
04ML tooling
Stable · v2.0Michi
道 — a local-first ML workbench. Independent CLI tools that profile datasets, evaluate models, and benchmark them with real statistical rigor. Automates the implementation, never the judgement.
- independent CLI verbs
- 18
- leakage overstatement caught
- 26 pts
- network calls, fully local
- 0
Also built
More projects
Smaller ML and applied-AI work — experiments, prototypes, and deployments that shaped the flagship projects.
ANTIDOTE — 1st Place, IndiaCodex '26
Hackathon-winning agent knowledge-recall system (Masumi "Monetize AI Agents" track, Cardano): forged data is traced to every agent that ingested it and revoked, with 3 Aiken validators enforcing quarantine and Merkle non-membership proofs for provable deletion.
TypeScriptAikenReactMerkle proofs
Image Similarity Search
A high-performance image similarity search engine: MobileNetV2 embeddings in PyTorch, a custom k-d tree written in C for nearest-neighbor search, and an interactive web UI with 2D/3D embedding visualizations.
CPyTorchk-d treeD3.js
Reflective Reasoning Transformer
A research prototype exploring causally-grounded reasoning pretraining: training a language model on causal reasoning graphs, not just text, aiming at provably more faithful step-by-step reasoning.
PythonTransformersReasoning
AI Medical Report Explainer
A biomedical NLP application: OCR over lab reports, NER-based lab-value extraction, chart generation, and LLM-generated plain-language explanations — an early exploration of the problems Meridian now takes on seriously.
PythonOCRNERBiomedical NLP
Movie Sentiment Analyzer
Cross-runtime ML deployment: a custom-trained PyTorch sentiment model exported to ONNX and served from a Java application, with results persisted to MySQL.
JavaPyTorchONNXMySQL