Rohith Shinoj Kumar
Scientist 'B' @ C-DOT (Govt of India)
I am a systems researcher and software engineer specializing in high-throughput C/C++ architecture, DPDK kernel-bypass packet engines sustaining 400 Gbps line rates, and lock-free concurrency. A Computer Science graduate from NIT Surathkal, my work spans national-scale cyber-infrastructure at C-DOT, high-performance AI/LLM inference stacks, and quantitative time-series analytics using kdb+/q.
Work Experience
Education
National Institute of Technology, Surathkal
B. Tech. in Computer Science and Engineering
Asian International Pvt School, Abu Dhabi, UAE
AISSCE Class XII (CBSE Board)
Featured Projects
Finugreek — Market Analytics & AI Quant Execution Platform
Zero-Downtime High-Frequency Tick Ingestion & AI Quantitative Execution Engine
- Engineered an ultra-low latency time-series ingestion pipeline leveraging kdb+/q to parse high-frequency tick and L2 order-book data via persistent WebSockets.
- Implemented vectorized query execution and in-memory columnar aggregations for dynamic factor models and multi-asset correlation matrices at sub-millisecond latency.
- Architected a concurrent data normalization pipeline with a 32-process worker pool compiling compressed Parquet column-stores for DuckDB zero-copy OLAP analytics.
- Deployed a 24/7 cloud backend on Oracle Cloud Infrastructure (OCI) with Docker, securing FastAPI SSE streaming endpoints via Cloudflare Zero Trust Tunnels.
- Orchestrated autonomous LangGraph AI research agents delivering real-time factor attribution and portfolio strategy insights.
Local LLM & Cortex MCP Context Engine
High-Throughput Local Inference Stack with Tree-Sitter Codebase Distillation
- Architected and deployed a local vLLM inference stack hosting Qwen3.6-27B (FP8) with RadixAttention KV-cache prefix caching, achieving ~36 tok/s throughput.
- Built a hybrid RAG & MCP infrastructure using Qdrant vector DB with BAAI/bge-m3 1024-dim embeddings integrated via FastMCP for dynamic file selection and web search.
- Engineered Cortex, a native custom CLI Context Engine using Tree-Sitter AST parsing and Watchdog daemons to generate compressed codebase maps.
- Achieved 90% token reduction in context window utilization while maintaining complete local code confidentiality and zero data egress.
Reliable, Auditable Messaging Platform for Regulated Industries
Fault-Tolerant High-Availability Messaging Backbone with Guaranteed Delivery
- Designed and implemented a scalable, fault-tolerant messaging backbone requiring guaranteed delivery, end-to-end audit trails, and zero data loss.
- Leveraged Kafka for durable event streaming paired with Redis as a low-latency replay buffer and dead-letter queue system.
- Structured automated audit logging to PostgreSQL with ACID transactional safety under high concurrent load.
- Built a fully dockerized deployment stack including Kafka, Redis, Postgres, and FastAPI app server for reproducible local and cloud environments.
Technical Skills & Core Competencies
Low-Latency & Systems
AI & LLM Infrastructure
Networks & Security Protocols
Cloud, Containers & DevOps
Programming Languages
Machine Learning & Analytics
Certifications & Honors
kdb+/q Developer Certification
KX
Chartered Financial Analyst (CFA) Level-I
CFA Institute
Google Cybersecurity Specialization
AWS Fundamentals
Amazon Web Services
UAE Golden Visa (Academic Performance Honors)
Government of UAE
Publications
Automated Rib Segmentation in Chest X-rays Using ThoraxSegNet: Enhancing Pulmonary Disease Detection and Analysis
IEEE Access
2026
A Deep Learning Framework for Automated and Consistent Ejection Fraction Quantification in Echocardiography
International Journal of Biomedical Imaging
2026
H-Infinity Filter Enhanced CNN-LSTM for Arrhythmia Detection from Heart Sound Recordings
15th IEEE International Conference on System Engineering and Technology (ICSET), Kuala Lumpur, Malaysia
September 2025
Deep Learning based framework for dynamic Detection and Mitigation of ARP Spoofing attacks
14th IEEE Conference on Computing, Communication and Networking Technologies (ICCCNT), IIT Delhi
July 2023