Headline

Data Science Engineer at Evertz Microsystem | Genearative AI | Agentic AI | MLOPs | Machine Learning | Deep Learning | NLP | Data Engineering | Data Analytics

About

I’m Shubham Kokila, a Data Science Engineer with hands-on experience in GenAI, AgenticAI, Machine Learning, Deep Learning, and MLOps. Currently working at Evertz Microsystem, I have led impactful projects ranging from log anomaly detection using BiLSTM and LLMs, to building a RAG-based chatbot (EVGPT). My expertise spans the entire ML lifecycle—from data ingestion and model training to deployment and monitoring using tools like ZenML, MLflow, Optuna, and Aimflow. With a strong academic background, I hold a B.E. in Computer Science (CGPA: 9.0) from Acharya Institute of Technology and am concurrently pursuing a B.S. in Data Science from IIT Madras. I’ve also built practical applications like sentiment analysis models and full-stack web apps using Flask and PostgreSQL, reflecting my versatility in both data science and software development. Passionate about GenAI, I’ve worked with Hugging Face models such as Falcon and Zephyr, integrating them with tools like ELK stack and Chainlit for robust solutions. I thrive in collaborative environments and bring a mix of technical skill, creativity, and a drive for continuous learning to every project I undertake.

Location

Bangalore Urban

AI stack

claude code

grok

hugging face

langchain

langfuse

openai

Work experience

Data Science Engineer 1 at Evertz India Pvt Ltd, May 2026 - Present · 4 mos, Bengaluru, Karnataka, India · Hybrid, 𝗶𝗻𝗦𝗜𝗧𝗘 𝗗𝗲𝗽𝗹𝗼𝘆𝗺𝗲𝗻𝘁 | 𝗘𝗟𝗞 𝗦𝘁𝗮𝗰𝗸 | 𝗔𝗻𝗮𝗹𝘆𝘁𝗶𝗰𝘀 | 𝗚𝗲𝗻𝗔𝗜 □ Spearheaded deployment and integration of Evertz’s inSITE (ELK Stack-based big data analytics platform) for enterprise and global broadcast clients. □ Configured and optimized ELK Stack components including Logstash, Elasticsearch, and Kibana for centralized log management, monitoring, and analytics workflows.

Junior Data Science Engineer at Evertz India Pvt Ltd, Aug 2024 - May 2026 · 1 yr 10 mos, Bengaluru, Karnataka, India · Hybrid, 𝗔𝗜/𝗠𝗟 𝗦𝘆𝘀𝘁𝗲𝗺𝘀 | 𝗠𝗟𝗢𝗽𝘀 | 𝗜𝗦𝗥 𝗩𝗶𝗱𝗲𝗼 𝗔𝗻𝗮𝗹𝘆𝘁𝗶𝗰𝘀 | 𝗚𝗲𝗻𝗲𝗿𝗮𝘁𝗶𝘃𝗲 𝗔𝗜 𝗲𝘃𝗜𝗦𝗥 𝗔𝗻𝗮𝗹𝘆𝘁𝗶𝗰𝘀 – 𝗥𝗲𝗮𝗹-𝗧𝗶𝗺𝗲 𝗜𝗦𝗥 𝗩𝗶𝗱𝗲𝗼 𝗜𝗻𝘁𝗲𝗹𝗹𝗶𝗴𝗲𝗻𝗰𝗲 □ Developed AI-powered ISR video analytics pipelines for real-time object detection, tracking, and inference on 30+ FPS video streams. □ Benchmarked and optimized RF-DETR variants, improving detection mAP by 12% while reducing inference latency by 25%. □ Built ReID-based tracking workflows that reduced ID-switching errors by 35% across dense surveillance scenes. □ Implemented Kafka streaming pipelines supporting 10+ concurrent camera feeds for scalable real-time analytics. □ Integrated MinIO-based inference storage handling 100K+ structured inference outputs efficiently. □ Optimized GPU inference workflows, improving throughput by 40% and reducing response latency by 30%. 𝗟𝗼𝗴 𝗔𝗻𝗼𝗺𝗮𝗹𝘆 𝗗𝗲𝘁𝗲𝗰𝘁𝗶𝗼𝗻 – 𝗠𝗟𝗢𝗽𝘀 𝗣𝗹𝗮𝘁𝗳𝗼𝗿𝗺 □ Developed an end-to-end MLOps pipeline for log anomaly detection using BiLSTM models integrated with LLM-generated incident summaries. □ Reduced false positives in production log anomaly detection by 28% through model optimization and tuning. □ Built automated ZenML pipelines for ingestion, training, evaluation, deployment, and monitoring workflows. □ Improved experiment tracking and reproducibility using MLflow and Aim, increasing monitoring efficiency by 40%. □ Integrated Optuna hyperparameter optimization with pruning strategies, reducing tuning time by 45%. □ Automated AI-generated incident response summaries, reducing manual log analysis effort by 30%. □ Leveraged LangChain and LangGraph for intelligent AI workflow orchestration and operational analysis.

Student Engineer at Evertz India Pvt Ltd, Jan 2024 - Jul 2024 · 7 mos, Bengaluru, Karnataka, India · Hybrid, 𝗘𝗩𝗚𝗣𝗧 (𝗥𝗔𝗚-𝗯𝗮𝘀𝗲𝗱 𝗘𝗻𝘁𝗲𝗿𝗽𝗿𝗶𝘀𝗲 𝗖𝗵𝗮𝘁𝗯𝗼𝘁) | 𝗚𝗲𝗻𝗲𝗿𝗮𝘁𝗶𝘃𝗲 𝗔𝗜 □ Researched and developed a Retrieval-Augmented Generation (RAG) chatbot prototype for enterprise knowledge assistance. □ Integrated PostgreSQL vector storage with open-source LLMs including Mistral, Falcon, and Zephyr via Hugging Face. □ Built interactive chatbot interfaces using Chainlit, improving user experience and reducing query response time by 40%. □ Designed scalable GenAI workflows for efficient enterprise document retrieval and contextual response generation. 𝗘𝗟𝗞 𝗦𝘁𝗮𝗰𝗸 | 𝗟𝗼𝗴 𝗔𝗻𝗮𝗹𝘆𝘁𝗶𝗰𝘀 | 𝗔𝗜-𝗽𝗼𝘄𝗲𝗿𝗲𝗱 𝗤𝘂𝗲𝗿𝘆𝗶𝗻𝗴 □ Developed ELK Stack pipelines using Elasticsearch, Logstash, and Kibana for centralized log ingestion and visualization. □ Automated data collection using Filebeat and Winlogbeat, with interactive Kibana dashboards for operational monitoring. □ Implemented a RAG-based chatbot over Elastic vector stores for intelligent log querying, improving log retrieval efficiency by 35%.

Education

Diploma, Data Science and Application, Indian Institute of Technology, Madras

Bachelor of Engineering - BE, Computer Science, Acharya

Public links

github: https://github.com/shubhamkokila