Aditya Kishtawal

Led the architecture and development of 45+ production-grade agentic AI systems as the first AI engineer on the founding team, processing 50M+ insurance records for automated carrier mapping and document intelligence.

Achieved 45.7% performance optimization in large-scale data pipelines by engineering async architectures and distributed caching, saving ~1000 days in processing time for real-time business intelligence.

Developed multimodal document intelligence systems for SERFF regulatory filings using Vision LLMs, LangChain, and Pinecone, automating structured data extraction and compliance analysis across 50+ US states.

Built compound AI systems, achieving 80-90% autonomous coverage for carrier-to-business mapping and NAICS classification, utilizing ML-based firmographic regression models and clustering methodologies.

Deployed robust production monitoring infrastructure using Streamlit, PostgreSQL, and AWS, providing real-time dashboards for system performance and data quality across all AI pipelines and client-facing applications.

Scaled AI engineering operations by 125% (from 4 to 9 members), establishing technical hiring protocols and conducting 25+ candidate interviews to secure 5 strategic hires.

Mastered Data Analytics, Statistics, Machine Learning (MLOps), Deep Learning, and NLP, gaining comprehensive theoretical and hands-on Data Science expertise.

Led the end-to-end development of 3-5 GenAI-based client MVPs, leveraging cutting-edge LLM models in Python, prompt engineering, and solution architecture building.

Developed and deployed innovative AI solutions, including an Automated Course Development Model, Voice-based Search Enhancement for E-Commerce, and YourHonor-AI for the Indian Legal System.

Applied advanced GenAI concepts and technologies, including Retrieval Augmented Generation (RAG), Vector Databases (FAISS, Pinecone), and Fine-tuning, to enhance project outcomes and mitigate risks.

Leveraged Python development and Data Analytics skills to automate data extraction capabilities for Zintlr's B2B lead generation platform.

Collaborated with cross-functional teams to develop and deploy automated Django APIs and data scraping scripts, optimizing lead generation processes and increasing overall leads by 20-30%.

Led the complete development of a 'LinkedIn Parser' for Zintlr's Chrome extension, automating key information extraction and reducing parsing time by over 80% using Scrapy and XPaths.

Integrated parser scripts with MongoDB using PyMongo and created APIs for seamless front-end integration, streamlining data flow and ensuring data accuracy through rigorous testing.

Spearheaded a pivotal research project within the Atmospheric and Oceanic Sciences Group (AOSG) titled 'Bias Correction of SCATSAT-1 Winds' for ISRO's scatterometer satellite.

Steered data collection and analysis efforts, collating over 100 GB of diverse datasets using Python, and identified strategic opportunities to enhance satellite data quality and communication accuracy.

Researched, applied, and compared pertinent data science and machine learning algorithms to rectify bias in satellite wind data, achieving respective accuracies ranging between 75-90%.

Presented research findings to the esteemed scientific peer group at AOSG, receiving validation to proceed with documenting the research.