• Engineered advanced agentic AI systems leveraging tool and function calling to orchestrate complex, multi-step workflows, integrating advanced LLMs like OpenAI, Anthropic, and Mistral to power both parallel and sequential task execution.

• Optimized machine learning models using TensorFlow and Keras, enabling live data processing and achieving a 20% reduction in inference time for real-time user queries.

• Single handedly researched, developed, and iteratively refined 70+ prompts for intent-entity recognition, and contextual query handling, enhancing response quality by 30% for EVA (Enterprise Virtual Assistant), the next-gen AI product.

• Orchestrated data flow using REST APIs for seamless communication between EVA’s agents and external systems like Jira, HubSpot, and Zendesk, supporting over 1.2 million API calls per month with a 99.9% uptime.

• Maintained user level session using Redis in order to support follow up questions from the user.

• Developed a resume parser from scratch using Python and NLTK to extract key details (name, email, skills).

• Deployed a Node.js client-server model with Amazon S3 integration, processing 100+ resumes daily with scalable cloud storage.

• Championed the innovation of an AI-driven optimization method for Multi-Camera Positioning using Python, achieving a 25% improvement in installation efficiency by streamlining the setup process.

• Designed an advanced algorithm leveraging camera lens parameters, and computer vision for precise object localization, utilizing particle swarm optimization to minimize camera usage while ensuring optimal area coverage and video quality.

• Won a Certificate of Excellence for outstanding leadership in project execution, demonstrating strong team collaboration, technical expertise, and strategic communication skills.

• Developed ML models for global-scale application software, leveraging K-Means Clustering to analyze a dataset over 1 million words of customer request.

• Conducted in-depth text analysis and implemented unsupervised learning methods to streamline customer query classification, enhancing response categorization accuracy by 35%.

• Implemented financial data feeds by processing and transforming raw data for seamless integration with a web-based dashboard, reducing data inconsistencies by 25%.

• Created interactive data visualizations using Perspective, enabling real-time financial trend analysis with millisecond-level latency, thereby enhancing decision-making processes.