Yash Panchal

Developed SFDR Gen-AI Reporting application:

• Spearheaded the development and successful delivery of an advanced Gen AI-based reporting tool aligned with SFDR (Sustainable Finance Disclosure Regulation) guidelines.
• Designed and fine-tuned prompts to optimize KPI extraction and enhance the accuracy and relevance of AI-generated insights from sustainability reports.
• Acted as a critical link between cross-functional teams—including backend, frontend, and strategy—by articulating innovative solutions and ensuring seamless collaboration throughout the application development lifecycle.

Gen-AI driven Conversational chatbot for ESG360 application:

• Ideated, designed, and architected the ESG CoPilot Insights platform, seamlessly integrating it with an ESG-focused Large Language Model (LLM) to deliver customized client demos and empower sales enablement initiatives.
• Leveraged cutting-edge tools including LangChain, Llama Index, and GPT-based models to develop a conversational chatbot for the ESG360 tool, enabling the extraction of personalized, user-specific insights in real time.

GreenFInT – Financed and Insured Emissions Development

• Designed, optimized, and developed a comprehensive tool for calculating, visualizing, and tracking financed emissions per industry standards such as PCAF and SBTi. Integrated robust scenario modeling capabilities within the tool.
• Leveraged data analytics and Gen AI algorithms to automate Excel-based models. Applied functional expertise in both front-end and back-end development of a web-based tool for financed emissions accounting and visualization, as well as target setting for future forecasting.

Proficiency in Addressing Sustainability Challenges through Data and AI

• Leveraged data analytics and AI for insightful, sustainable decision-making, improving transparency and accountability in sustainability reporting

TEN-XR OneEdge – Edge Deployment of Deep Learning Object Detection Model

• Developed a object detection model with sub-second inference time, incorporating shelf status classification, and deployed it using Flask. Built Docker images for edge deployment and pushed them to Azure Container Registry (ACR).

Loss Forecasting Model:

• Developed a Loss Forecasting model by converting an MS Excel model to Python, optimizing runtime from 40+ minutes to under a minute. Worked with over 7 years of big data, incorporating ad-hoc changes and modifications.
• Forecasted losses for the next decade based on historical data and created an interactive Tableau Dashboard displaying model results with multiple filters for year, quarter, and MOB readability.

Crop Classification using Supervised Deep Learning and Unsupervised Model:

• Developed a classification model for identifying crops in specific areas (village, district, state) using multi-spectral raster images (.tif) from Sentinel satellite data over 6 months. Employed NDVI for vegetation assessment and applied classification models including Logistic Regression, 3-layer and 4-layer Neural Networks, and Random Forest for comparison.
• Expanded to unsupervised methods like K-means Clustering and Gaussian Mixture Model, conducting a multi-crop classification study in DHAR district, Madhya Pradesh.

• Designed and developed Mechanical Body and Structure of CHANDRAYAAN-II payloads, IIRS, TMC-2 and Rover-Image
• Qualified and approved design, assembly process, drawings of Packages, Sub-Assembly, Assembly and Payload.
• Designed specific application mechanical structures, for research purpose and Technology Development Process (TDP)