Open to opportunities

Pratik Karnik

@pratikskarnik

Message

I am an AI engineer working in Cognizant

India

Message

What I'm looking for

I want a Data Scientist job where in I can make autonomous decisions and have ownership of the product.

Experience

Cognizant | AI Engineer

Healthcare RAG: Built a QA system (Python, LangChain, PGVector), cutting manual document processing by 90%.
Agentic Automation: Led a Vertex AI Agent Builder POC for 5,000 users, enabling an 80% reduction in team size.
Knowledge Graphs: Integrated FalkorDB & OpenAI into an agentic framework in 4 weeks, boosting dev efficiency by 80%.
Auto-Docs: Developed automated architecture & README generation from KGs within 2 weeks.

TCS | Data Scientist

Industrial ML: Optimized oil production via DBSCAN clustering, driving significant cost savings.
Automation: Engineered Python scripts for automated SQL generation.

Searce Inc. | ML Engineer

Infrastructure: Spearheaded Vertex AI pipelines for scalable model deployment.
Computer Vision: Built a deep learning background removal app with high evaluation metrics.

LTI | Consultant (2018–2021)

Managed JDE ERP security, compliance, and regulatory support.

Experience

Work history, roles, and key accomplishments

Current

Sr. Associate - Projects

Current

Cognizant

Nov 2024 - Present (1 year 3 months)

Developed RAG-based healthcare chatbot; cut manual processing by 90%.

Built Vertex AI POC for 5,000 users; reduced team size by 80%.

Integrated FalkorDB Knowledge Graph in 4 weeks; boosted dev efficiency by 80%.

Python Generative AI Neo4j LangChain Libraries LangGraph GraphDb Cypher Query Tuning LLM OpenAI Models OpenAI SDK Claude Sonnet 3.5 Anthropic: Claude PostgreSQL Databricks Git Gitlab Gemini Lang Chain

Data Scientist

Tata Consultancy Services

May 2022 - Oct 2024 (2 years 5 months)

Developed an efficient algorithm to optimize resource allocation by assigning engineers to different sites. Automated data entry by developing Python scripts to enter SQL statements which reduced manual entry time by 80%. Utilized DBSCAN algorithm in scikit-learn package to accurately predict non maintenance intervention in oil production, resulting in a cost savings
of $5M per true positive

Python SQL Machine Learning Data Science NumPy Pandas TensorFlow Keras scikit learn MLOps

ML Engineer

Searce Inc.

Oct 2021 - May 2022 (7 months)

• Implemented Vertex AI training and inference pipelines using Python and Google Cloud Functions which get triggered on adding data to GCS buckets
• Trained and evaluated the model achieving an impressive R2 score of 0.79 on a 15-year test dataset
• Deployed a model capable of removing backgrounds from high-resolution images with dimensions of 864x864 using unet architecture.