Surya Pratap

Working as an AI Developer at Xicom Technologies, I have been working on building powerful Agentic AI, who are capable of a specific domain and mimic a real human using personification. Also I have worked on Multi-Agent Orchestration Solution, and more.

My role centers on creating innovative solutions to identify individuals across two distinct image sets. I leverage my expertise to craft software capable of analyzing still images from videos, accommodating various orientations and lighting conditions. This may involve web scraping for image collection. Adhering to provided tasks and deadlines, I collaborate closely with supervisors, ensuring clarity and addressing any queries promptly to deliver optimal outcomes.

During my one-year tenure at Let’s AI Solution as an AI/ML Developer, I designed and deployed cutting-edge AI solutions leveraging LangChain, Azure AI, and multi-agent frameworks such as AutoGen and CrewAI. I fine-tuned multiple LLMs and SLMs (including Llama 3B–8B, Phi 3/4, Ministral, and Google T5) using LoRA, PEFT, and supervised fine-tuning techniques to enhance model performance. My key projects included developing Picturesque, an AI-powered event management system integrating computer vision for facial recognition, automated video editing, and multimedia generation, and architecting a Tender Validation System powered by AutoGen agents that reduced validation time from days to minutes with 95% accuracy. Additionally, I built a call center analytics platform integrating OpenAI Whisper, Azure AI, and custom scoring algorithms for transcription and conversation insights. Working in a dynamic startup environment, I independently managed end-to-end development cycles, showcasing strong technical expertise, problem-solving, and leadership skills.