loading
reading
writing
saving
searching
AI / ML Engineer | Computer Vision, RAG, Deep Learning & Production AI
QR Code Link to This Post
I'm an AI / machine learning engineer, and most of my work is about building systems that don't just look good in a notebook, but actually hold up in production.
I've worked across computer vision, deep learning, RAG, vector search, and ML infrastructure. A lot of what I enjoy is the engineering side of AI — making models faster, reducing wasted compute, improving retrieval quality, and turning ideas into something practical.
COMPUTER VISION / OBJECT DETECTION
One area I've spent a lot of time on is computer vision.
In one project, I built a multi-stage detection pipeline where a smaller model handled the easy cases first, and only passed uncertain regions to heavier models. That helped cut a lot of unnecessary compute, and the system ended up running about 2x faster than the baseline without a meaningful hit to accuracy.
I've also worked on retrieval-based approaches for visual tasks, where similar examples are pulled from an indexed dataset and used to improve how edge cases are handled.
My computer vision work includes:
- Object detection
- Image classification
- Real-time inference pipelines
- Custom PyTorch models
- Model optimization
- High-throughput vision systems
RAG / VECTOR SEARCH / RETRIEVAL
I've also built a lot of the infrastructure behind retrieval systems.
That includes document pipelines, embedding generation, vector indexing, semantic search, and integrating retrieval into AI applications. I've worked with FAISS and Pinecone, including retrieval setups in the 10M+ vector range, where latency and recall both matter.
What I care about here is not just making RAG work, but making it usable in production — fast enough, accurate enough, and maintainable.
My experience includes:
- RAG pipelines
- FAISS / Pinecone
- Embedding generation
- Semantic search
- Document ingestion and indexing
- Retrieval optimization
- Knowledge search systems
- LLM and retrieval integration
PREDICTIVE MODELING
I've also built predictive systems in areas like crypto forecasting and sports prediction.
A lot of that work involved using deep learning models for the prediction side, then pairing them with genetic algorithms to explore hyperparameters and strategy spaces that are hard to tune well with standard gradient-based methods alone.
Those projects taught me a lot about real-world data problems — noisy signals, leakage, unstable patterns, and how quickly performance can break when the environment shifts.
PRODUCTION AI ENGINEERING
I'm especially interested in the gap between 'the model works' and 'the system works.'
That usually means things like:
- Building inference APIs
- Designing real-time or batch ML pipelines
- Improving latency
- Integrating models into real products
- Scaling retrieval systems
- Debugging weak points in existing ML systems
TECH STACK
ML / AI:
Python, PyTorch, CNNs, Transformers, LSTM, DEAP
Computer Vision:
Object Detection, Image Classification, Real-Time Inference
Retrieval / Search:
FAISS, Pinecone, Embeddings, RAG, Semantic Search
Backend / Infrastructure:
FastAPI, Django, Node.js, REST APIs, PostgreSQL, Docker, Linux
WHAT I'M LOOKING FOR
I'm open to:
- Freelance AI / ML work
- Contract engineering
- Computer vision projects
- RAG and retrieval system work
- Model optimization
- Production AI backend development
- Technical collaborations
If you're working on something interesting and need help building or improving an AI system, feel free to reach out.
If you message me, it helps if you include:
1. A short description of the project
2. What problem you're trying to solve
3. Your current stage
4. Your budget and timeline
I'm open to remote work and longer-term collaborations.
Feel free to contact me through Craigslist email or chat.