Software Engineer · Systems + ML · Open to Roles
I'm Dhir Katre — a software engineer who builds across the full stack: distributed systems, cloud infrastructure, and applied machine learning. I design and train ML models, deploy them to production, and write the systems-level code that makes everything run at scale. M.S. CS at Oregon State.
I'm a software engineer who doesn't specialize narrowly — I build across systems programming, backend infrastructure, cloud platforms, and applied machine learning. The breadth is intentional: the best software at the intersection of these disciplines requires someone who speaks all of them fluently.
On the ML side, I design and train models — Mamba State Space Models, Reinforcement Learning (IQL, Decision Transformer), collaborative filtering with SVD via Surprise, and semantic search with Sentence Transformers + FAISS. I export to ONNX and build inference pipelines that actually run fast in production — not just in notebooks.
On the systems side, I write Rust (async/Tokio, WASM sandboxing), Java (Spring Boot microservices), and Python (FastAPI, SQLAlchemy, data pipelines). I've shipped cloud infrastructure on GCP, Azure, and AWS, and I care deeply about observability, correctness, and performance at every layer.
Machine Learning & AI
Cloud & Infrastructure
Backend & APIs
Languages & Systems
Feb 2024
Jul 2024
CVoter News Services
Jan 2023
Jun 2023
Ignite LLC
Production ML inference pipeline embedded in a CME futures market-making engine. Trained a Mamba State Space Model (10-50× faster than Transformers) in PyTorch, exported to ONNX, and served via Rust ONNX Runtime achieving sub-800ns end-to-end inference. Includes IQL and Decision Transformer for offline RL market-making, event-driven backtesting with realistic latency simulation, and real-time Prometheus/Grafana monitoring dashboards.
Asynchronous WebAssembly execution sandbox for safely running untrusted code in
isolation — the kind of infrastructure needed for secure ML model serving.
Custom ResourceLimiter enforces memory caps and CPU fuel limits
(instruction-level metering) to prevent DoS. Sub-microsecond structured telemetry
tracks execution duration and consumption metrics.
Published CLI tool with 200+ active users on Crates.io. Detects unused dependencies, functions, and modules via sub-ms AST traversal. JSON output for CI/CD.
Scalable identity provider microservice — Java 21 + Spring Boot 3.2. JWT auth, BCrypt hashing, stateless security filters, MySQL with strict FK integrity.
Collaborative filtering + content-based ML recommendation engine. Scikit-learn, user ratings + metadata, with explainable similarity scores.
ZK age verification on Ethereum. Proves age requirements without revealing identity — privacy-preserving compliance using cryptographic proofs.
Decentralized voting on Ethereum. Tamper-proof ballot recording and transparent vote tallying without centralized authority.
High-performance Binomial Options Pricing Model in Rust. Zero-cost abstractions and cache-friendly data structures for maximum quant throughput.
Concurrent reservation backend with race condition prevention, transactional seat locking, and high-throughput design.
Full-stack library management — book cataloging, member management, and loan tracking. REST API + frontend for staff operations.
M.S. Computer Science
Computer Science
B.E. Computer Science
Computer Science (Cyber Security)
Patent · India Patent Office
Automated Resume Analysis and Template Generation System
I'm actively looking for software engineering roles across backend systems, cloud infrastructure, and applied ML — internships and full-time. If you're building something technically challenging, I'd love to talk.