Shreyas S K
Senior Agentic AI Engineer
6+ years of experience designing production-grade LLM systems, RAG pipelines, and agentic workflows across enterprise environments. Proven track record leading cross-functional initiatives from architecture to deployment.
Core Technical Expertise
Programming
Python, PyTorch, Numpy, Pandas, Langchain, Langgraph
AI Technologies
Machine Learning, NLP, Computer Vision, Reinforcement Learning
Cloud Platforms
AWS, GCP: ECS, EC2, S3, Sagemaker, Cloud Build, Cloud Run
Generative AI
OpenAI, Gemini, Llama, Phi3, DeepSeek, Gemma, LoRA, QLoRA, RLHF, DPO
MLOps/LLMOps
Dagster, Argilla, Deepeval, Opik, Docker, Kubernetes, WandB, DeepSpeed, vLLM, accelerate, Nvidia Triton, TensorRT, Qdrant, Pixeltable
Teradata Corporation
Senior Agentic AI Engineer | Nov 2025 - Present
OneTeradata Agent
Leading innovation in multi-agent orchestration. Implementing a sophisticated framework enabling customers to perform data exploration, metadata extraction, help desk operations, database stats, and formatting through natural language interface.
This groundbreaking work transforms how users interact with complex database systems, making enterprise data management accessible through conversational AI.
4
Agentic AI Capabilities
SLM/LLM Finetuning
5
Core Functions
Natural language capabilities
Ford Motors Pvt Ltd
Generative AI Specialist | Jul 2024 - Oct 2025
AI DevOps Agent
Led team of 5 building AI-powered DevSecOps automation platform integrating Terraform, CI/CD, and security compliance
Customer Impact
Onboarded 150+ internal customers with natural-language deployment and compliance automation
Efficiency Gains
Achieved 30-60% reduction in engineering effort through pipeline automation
In-Vehicle Assistant
Developed multi-agent system handling vehicle controls, navigation, entertainment, and information retrieval. Evaluated 15+ LLMs on function calling accuracy, achieving ~20% improvement through fine-tuning Llama 3.2 and Qwen 2.5 3B models. Deployed on Nvidia H100 with Triton and TensorRT optimizations achieving <50ms latency.
30-60%
Effort Reduction
Engineering efficiency improvement
20%
Model Performance
Improvement through fine-tuning
50ms
Latency
Response time achieved
Khoros India R&D Pvt Ltd
Machine Learning Engineer | Jan 2022 - Jul 2024
Helix RAG-Powered Bot
Built analytics engine using Taxonomy/Ontology and Topic Modeling orchestrated with Dagster. Extracted entities as metadata for fine-grained answers, fine-tuned retriever for domain terminology and abbreviations, and optimized reranker for answer relevancy.
Reduced agent handoff rate by 38% through improved retrieval precision and content routing.
Agent Auto Assist
Fine-tuned Mistral 7B on 1.5M call-center conversations to power smart compose functionality. Collected accept/reject feedback and applied DPO for preference alignment, maintaining brand tone and compliance standards.
38%
Handoff Reduction
Improved retrieval precision
61%
Response Time
Faster agent responses
Meta: AR/VR Innovation
ML Engineer (HCL Client Project) | Apr 2020 - Jan 2022
1
Test Automation CLI
Developed for Facebook's AR/VR/VA products including Oculus and Ray-Ban Stories. Implemented image binarization with SciPy/Numba cutting OCR compute by 97%.
2
Augmented Reality Effects Detection
Deployed Detectron2 model for AR effects detection in WhatsApp/Messenger video calls. Owned complete data pipeline, training, and deployment process.
3
Textual Similarity
Delivered DistilBERT-based textual similarity with high accuracy and low latency. CLI reduced execution time by 50%.
97%
Compute Reduction
OCR optimization achievement
50%
Time Savings
CLI execution improvement
HCL Technologies
Lead Engineer | Jul 2019 - Jan 2022
COVID-19 Classification
Modified DenseNet-121 achieved 96% accuracy for X-ray classification. Quantized for edge deployment with 50-75% size reduction and minimal accuracy loss.
Mobile GPU Optimization
Android app to benchmark ARM Mali GPU inference. Improved GPU inference speed by 30% (FP32) and 80% (FP16) via TFLite optimization.
96%
Model Accuracy
COVID-19 detection
50-75%
Size Reduction
Edge optimization
80%
Speed Boost
FP16 inference
Education & Continuous Learning
IIT Guwahati
M.Tech | Aug 2017 - Jun 2019
CGPA: 9.79/10
🏆 1st place in merit list
P.E.S University
B.Tech | Jul 2013 - May 2017
CGPA: 9.14/10
🏆 7th place in merit list
IISc Bangalore
Specialized Courses
A+ in AI for Medical Imaging
A in Deep Reinforcement Learning
Coding Ninjas
Content Reviewer
Designed comprehensive Generative AI course syllabus for non-technical professionals across HR, Sales, Marketing, L&D, and Business Analytics.
NCERT, Ministry of Education, Government of India
Member of Textbook Dev Team
Contribute to Grades 11 & 12, Foundations and Methods of Artificial Intelligence textbooks aligned with NEP 2020 and NCF 2023
Guvi Geek Networks
Content Reviewer
Reviewed YouTube video modules for Generative AI course in Kannada, ensuring accessibility and technical accuracy for local learners.
Research Publications & Impact
01
Edge-AI-IoT (2021)
Shreyas and Kameshwar. "Diagnostic Decision Support for Medical Imaging and COVID-19 Image Classification on ARM Mali GPU". IEEE Global Communications Conference 2021. Published December 7, 2021.
02
Infrastructure Solutions (2019)
Shreyas and Dey. "Application of soft computing techniques in Tunneling and Underground excavations: State of the art and future prospects". Innovative Infrastructure Solutions, Springer Publications. Published September 25, 2019.
03
IACMAG Symposium (2019)
Shreyas and Dey. "Comparative Efficacies of Ensemble methods and Hybrid Intelligent model in Predicting Penetration Rate of Tunnel Boring Machine". International IACMAG Symposium, IIT Gandhinagar. Published March 7, 2019.
Research contributions span medical imaging, infrastructure engineering, and machine learning applications, demonstrating versatility across domains.
Achievements & Recognition
1
Bengaluru Mobility Hackathon 2024
Finals Selection out of 470 teams. Organized by IISc, IUDX, BTP. Developed vehicle re-identification across multiple cameras for O-D flows.
2
SEBI Hackathon 2023
3rd Place out of 1200 teams at Global Fintech Fest. Developed AI model to detect misleading claims by financial influencers using Generative AI.
3
Autonomous Agents Hackathon
Top 25% out of 400 teams. Lablab.ai and Weaviate collaboration demonstrating advanced agent capabilities.
4
WhatFix Generative AI Hackathon
4th Place out of 60 teams. Contributed to AI solution for social good using Generative AI with AWS.
Vista Hackathon
Special Mentions for Content Genie - all-in-one social media post creation tool
Facebook Project Recognition
ERS Innovation Champion and ERS Business Champion for outstanding performance