About

Hello! I’m Rishikesh Ajay Ksheersagar, an AI Engineer and Data Scientist focused on building reliable, scalable LLM, RAG, and multimodal AI systems. I am currently working as a Data Scientist at LatentView Analytics, partnering with HP Inc.’s Personal Systems Quality team to design, build, and productionize AI solutions—including agentic RAG, CV-LLM compliance automation, support-case classification and summarization, and PySpark telemetry pipelines monitoring millions of devices. My work spans the complete AI engineering lifecycle, from system architecture and retrieval to deployment, observability, and latency optimization.

I recently completed my Master’s in Data Science at the University of Michigan, Ann Arbor (May 2025).

My journey in Data Science space began with a Bachelor’s in Computer Engineering from Savitribai Phule Pune University (May 2019).

Post my Bachelor’s, I spent four years at Mu Sigma Inc. (July 2019 - June 2023), as a Data Scientist and later as an Apprentice Leader (Data Science Manager), honing technical skills in AI/ML, Statistical Analyses, and Business Intelligence, as well as leadership and mentorship qualities. Check out my Work Experience.

At UofM, my research at the LAUNCH Lab under Dr. Lu Wang explored the evaluation and safety of agentic large language models, with a particular focus on detecting and stress-testing deceptive or scheming behavior in tool-augmented LLM systems. Further, my research at the Language and Information Technologies Lab, under Dr. Rada Mihalcea and Dr. Veronica Perez-Rosas, focused on NLP and Generative AI, particularly modeling fake news perception and analyzing how users interpret and respond to misinformation. Additionally, I contributed to the “Climate Change, Demographic Shifts, and Socio-Political Stability” project, developing NLP pipelines to analyze large-scale research corpora and identify causal relationships between climate change and social unrest. For more, visit my Research.

I also served as a Graduate Student Instructor for the QMSS 301 course at University of Michigan in the Fall 2024 and Winter 2025 semesters. For more details, visit my Teaching.

To explore what I’ve built hands-on, from Register-Augmented LLMs to GPTQ-Quantized+LoRA fine-tuning pipelines, head over to my Projects.

In my free time, I have been working on 2 projects:

• Building a large-scale evaluation framework using AutoGen to investigate scheming behavior in LLMs. This involves generating high-stakes, agentic software engineering scenarios and dynamically invoking thousands of real-world tools and APIs to test LLM safety and tool-use alignment under pressure.
• Exploring new methods of QAT for LLMs which can be effectively coupled with Hardware for efficient Quantization Aware Training.

I am also passionate about environmental and ecological initiatives, and have contributed through volunteering work involving data analysis and digital strategy with EcoServants. For more info check this.

I believe in the power of continuous learning and would love to talk about Machine Learning, AI / LLMs, and Data Science. Let’s connect and explore what we can build together!