Jaykumar Kotiya - AI Engineer & LLM Expert
Experienced AI/ML Engineer with 6+ years building production-grade machine learning systems across healthcare, finance, legal, and industrial domains. Specializing in LLMs, Generative AI, and RAG pipelines.
Education & Certifications
- Relevant Industry Certifications
Technical Expertise
LLMs & GenAI
GPT-4, Gemini, BART, T5, Pegasus, LangChain, LoRA, RAG
Machine Learning & DL
XGBoost, CNNs, LSTMs, Transformers, Autoencoders, RL (PPO, DDPG)
MLOps & Cloud
FastAPI, Docker, Kubernetes, AWS, Azure, GCP, Snowflake
Proven Results
98%
Model Accuracy
Consistently delivering high-performance AI systems
<1s
Inference Latency
Fast, responsive AI solutions for real-time applications
60%
Workload Reduction
Automating manual processes with intelligent systems
99.5%
System Uptime
Enterprise-grade reliability for mission-critical ML systems
Clinical AI Innovation
At CitiusTech, I architected a GenAI clinical scribe using GPT-4 & LangChain for real-time medical transcription and summarization. This solution reduced physician documentation time by 60% and increased EHR note accuracy to 98%, streamlining patient care workflows.
RAG & Conversational AI
Knowledge Integration
Developed GPT-4-powered RAG chatbot with Pinecone & LlamaIndex for context-aware enterprise search
Business Impact
Boosted conversion by 22% and increased average session duration by 45 seconds
Performance
Improved answer relevance by 48% with Neo4j Graph DB integration and 800ms latency
Financial Intelligence
Built a real-time anomaly detection engine with LSTM + Transformer ensembles (PyTorch), ingesting live market data via Kafka. This system achieved 92% precision within 500ms, reducing trading risk and cutting financial losses by 35%.
Fraud Detection & Security
Engineered fraud detection microservices using XGBoost, autoencoders & AKS, enabling real-time transaction risk scoring with 99.5% system uptime and 28% fewer false positives under high-load conditions.
Implemented federated learning systems across 50+ edge devices for secure, decentralized training while ensuring zero data leakage.
Digital Twins & Reinforcement Learning
Agricultural Innovation
Engineered RL-based digital twin models using PPO and DDPG in PyTorch to simulate precision agriculture strategies—boosting simulated crop yield by 18%.
Geospatial Analysis
Built anomaly detection using YOLOv5, OpenCV, and QGIS to analyze satellite-tagged soil inputs with 92%+ accuracy.
Data Processing
Developed ETL pipelines with Apache Spark and Snowflake—cutting data processing time by 85% for agronomic decision-making.
MLOps & Automation Excellence
Pipeline Automation
Automated full MLOps lifecycle using Airflow and DVC—delivering reproducible, traceable pipelines from Snowflake ingestion to AKS deployment.
Efficiency Gains
Cut engineering workload by 60% through streamlined processes and intelligent automation.
Scalable Architecture
Designed systems handling 100K+ events/day with <1s latency and 99.5% uptime.
Contact Information
Languages
- English (Full Professional)
- Gujarati (Native)
- Hindi (Full Professional)
Education
MS, Information Systems - Northeastern University (2024)
BE, Electronics and Communication - Atmiya Institute (2018)