ML Engineer · Technical Writer
I specialize in production ML systems — inference optimization, memory management, and the gap between a model that works in a notebook and one that survives real load. I write about it too, with concrete numbers rather than hand-waving.
What I do
I work with teams and individuals who need machine learning expertise, clean Python code, or someone who can translate technical complexity into clear documentation and guides.
01
Inference optimization, training pipelines, and production deployment. Specializing in the part most projects underestimate: making models fast, memory-stable, and reliable under real load.
02
Clean, well-tested Python — data pipelines, automation scripts, CLI tools, and backend services built for maintainability.
03
Documentation, tutorials, and blog posts that make complex systems understandable — without losing accuracy.
04
Architecture review and design consultation for ML systems and data-intensive applications. Catch problems early.
Technical skills
Selected work
Production-grade projects built as engineering assessments — each one designed and tested as if it were going to production.
REST API for detecting and localising coins in images. Fine-tuned YOLOv8 Nano for detection, then applied deterministic geometric post-processing to derive precise bounding boxes, centres, and elliptical fits — avoiding the need for expensive pixel-level segmentation labels. Achieves 0.73 [email protected]:0.95 with 50–150 ms inference on CPU. Ships with 98% test coverage.
Production REST API for borehole image-log analysis. Ingests raw CSV image data, validates and anti-alias-resizes frames, persists them via a repository-pattern abstraction over SQLite, and serves depth-range queries with six domain-specific colormaps (resistivity, conductivity, geological, and more). The repository pattern keeps storage swappable — SQLite today, S3 or PostgreSQL tomorrow.
Writing
I write about production ML, inference optimization, and Python internals — on blog.rakeshraushan.org and Medium.
Let's work together
If you have a project in mind or just want to explore whether I'm a good fit — drop me an email. I typically respond within a day.