Hi, I’m Mark
I'm an ML Engineer at 314e Corp in Bengaluru, where I build and operate MLOps pipelines for healthcare ML systems — orchestration, distributed data processing, and model serving at scale. Outside of work I write low-level Rust and Python for fun: reimplementing NumPy from scratch, and building terminal tools I actually use every day.
Experience
Data Science Engineer · 314e Corp Current
- Architected and automated a full-scale MLOps pipeline for a patient entity-extraction model, raising accuracy from 72% to 94%; orchestrated fault-tolerant workflows with Temporal and deployed on SkyPilot, Hugging Face Endpoints, and RunPod with a human-in-the-loop refinement system.
- Re-engineered legacy Python scripts into scalable PySpark jobs for large-scale FHIR resources across multiple clients, cutting data-loading time by ~80%.
- Cut model training time from 90 hours to 25 hours on 500GB+ datasets via targeted hyperparameter tuning, and eliminated OOM errors with a custom memory-efficient iterable dataloader.
- Reduced model cold-start time by 77% by optimizing the serving process and pre-packaging assets into the runtime.
- Built a benchmarking suite for Vision-Language Models logged on ClearML, and reported a critical bug in Google's Gemma 3 VLM (Flash-Attention SDPA implementation).
Engineering writeups of this platform, which I contributed to — published by 314e, authored by Dr. Srivatsan Sridhar: AI document extraction & classification · IDP for medical records
Machine Learning Intern · Banach Technologies
- Engineered and optimized high-frequency trading strategies, reducing data-processing latency through targeted performance tuning.
- Built and deployed ML models from scratch against live market APIs to execute hedging strategies, backed by a comprehensive unittest suite for core trading libraries.
Software / ML Intern · Alemeno
- Built a data-preprocessing pipeline for semantic-segmentation models over large-scale GIS raster data, and implemented distributed computer-vision algorithms (Douglas–Peucker, Jarvis March).
- Deployed scalable Django apps on AWS with an end-to-end CI/CD pipeline, and hardened a high-performance raster-processing app through refactoring and unit tests.
Data Analyst · Bewgle
- Applied NLP techniques to proprietary Amazon-review datasets to deliver product-trend insights and predictive models; automated preprocessing in Python/Bash, cutting data redundancy by 90%.
Projects
A zero-dependency NumPy, written from scratch in pure Rust — same internals as the real thing (flat buffer + shape + strides). Reimplements broadcasting, axis reductions, and matmul from first principles, then benchmarks a cache-friendly
ikj loop order against BLAS-backed NumPy. Spec-driven with ~30 milestone tests.A terminal player that browses YouTube Music and your local library side by side, plays audio through
mpv over its JSON IPC socket, and scrobbles to Last.fm via cmusfm. Vim-style keys, live search, play history, and pixelated album art rendered straight in the terminal.Core ML/DL models — backpropagation, CNNs, transformers — implemented from first principles in pure PyTorch, without high-level abstractions.
An end-to-end LLM-powered resume parser with a modular OOP codebase and a Streamlit UI that turns an uploaded resume into structured JSON.
A CNN reaching 98.4% accuracy on medical waste classification, shipped as an interactive TensorFlow.js web app. Findings published in IEEE Xplore.
More on GitHub →
Skills
Languages
PythonC++RustBashJavaScript
Libraries & Frameworks
PyTorchTensorFlowscikit-learnNumPyPandasOpenCVRasterioMatplotlibDjangoFlaskFastAPIDashvLLMSGLangRunPod SDKSkyPilot
Tools & Platforms
DockerKubernetesAWS SageMakerMLflowClearMLTemporalGitLinuxPostgreSQLStreamlit
Education & Certifications
B.Tech, Computer Science — ABV-IIITM Gwalior, India · SGPA 8.67 · 2024
Relevant coursework: Artificial Intelligence, Statistics, Cloud Computing
DevOps on AWS Specialization · GCP Network Deployment · Open-source contributor @PyMC