Priyabrat Mishra

Responsibilities- • Contributing to the design & development of AI-native applications. • Building multi-agent workflows using LangChain, FastAPI, Python & supporting backend integration. • Researching SoTA models, latest AI tools & rapidly prototyping. Key contributions- • Building a Multi-Agent System (MAS) MVP for Employee attrition (targeting 85%+ accuracy) • Integrated RAG pipeline with FAISS & OpenAI embeddings (<1s latency, 85%+ recall) • Designed for 30% reduction in manual HR analysis time

1. Developed AI-driven tools using 4 CNN architectures, automating workflows with 90% accuracy. 2. Processed, cleaned, and annotated 120 images for FAIR-compliant datasets, boosting performance. 3. Optimized CNN models with hyperparameter tuning, improving metrics and accelerating training. 4. Used TensorBoard for logging, monitoring, and debugging model performance trends. 5. Integrated data from 3 simulation engines and 6 databases into GitLab CI/CD for AI analysis.

1. Revamped a SWIFT payment system data pipeline with four teams, reducing storage needs by 20%. 2. Managed data cleaning and verification from 12 sources, cutting processing time by 10%. 3. Developed a Confluence-based knowledge base, reducing team onboarding time by 30%. 4. Documented three modules using open-source Python tools, boosting developer productivity. 5. Built ETL server data pipelines, lowering release cycle times by 15%. Skills: Python | Data | Agile | JIRA | Confluence | Git | Kanban

1. Collaborated with 3 mainframe developers and 1 project manager on financial projects using IBM Mainframes. 2. Provided software support, ensuring seamless transactions and robust data integrity. 3. Developed a batch processing ATM system with 18 core functionalities, handling ~100 transactions per batch. 4. Reduced batch processing time by 15% through streamlined workflows and optimized code. 5.Created technical documentation for new mainframe features, accelerating onboarding for junior developers.