Fraction of my work that can be shared. Happy to share my work. Feel free to reach out to me if you need any assistance or have any questions.
Advanced drone vision system with trajectory planning, object detection, and visual SLAM.
Advanced 3D reconstruction system for creating detailed digital models.
- Build hardware of stereo vision from scratch
- Depth computation
- Custom stereo calibration
- Reading
End-to-end pipeline for multi-class segmentation and classification on prostate MRI data.
- Config-driven modular architecture with layered YAML composition
- Full HPC/SLURM support with sweep/grid search and auto model promotion
- Research runner orchestrating data import through inference
Foundation layer for agents over Postgres — agent-curated materialized views behind a two-role access boundary.
- Setup + runtime MCP servers; agent never touches base tables
- Access boundary enforced inside Postgres, not the app layer
- Lineage parsed by sqlglot; static HTML dashboard with refresh history
Agent-driven hyperparameter optimization with Optuna — Claude Code adapts the search strategy round-by-round.
- LLM agent reads round summaries via MCP, proposes search-space changes
- Optuna runs deterministically within each round (XGBoost, LightGBM, CatBoost)
- Auto-generated HTML reports per campaign; Postgres + MLflow backend
Agent-native CLI that discovers compute-heavy pandas operations and validates their Polars conversions.
- AST-based discovery with compute-weight ranking (loop-aware, impact-sorted)
- Validates correctness and benchmarks time/memory improvement
- Structured JSON output designed for autonomous agent loops
Event-driven inventory sync replacing daily batch pulls with near-real-time Shopify webhook updates.
- Shopify webhooks → FastAPI → Kafka (KRaft) → PostgreSQL pipeline
- Idempotent & order-safe with partition-keyed Kafka messages
- Fully containerized with mock webhook generator for local testing
Plain-text, git-tracked personal portfolio manager — check, visualize, and rebalance from terminal or dashboard.
- Drift-to-target in one command; EUR-native with auto FX conversion
- Trades live in CSV, targets in YAML; git is the audit trail
- Streamlit dashboard with inline editing; ~500 lines of Python
II. Publications
Deep Learning-Based Semiautomatic Generation of HD maps from Aerial Imagery
