Harshit Dave

I am a Data Scientist at slice, where I build production-scale LLM-powered agentic systems and multi-agent architectures for production applications These systems bridge cutting-edge research and real-world applications in the fintech space. My work is centered around LLMs, multimodal AI, and creating intelligent conversational systems.

My journey in AI/ML began at IIITM Gwalior, where I reseached and engineered the Integrated-Particle Swarm Optimization (i-PSO) technique and analyzed with various ML algorithms for detecting malware, under the supervision of Dr. Saumya Bhadauria and Prof. Aditya Trivedi. This early work taught me how to blend deep theoretical concepts with practical, real-world problem-solving.

Later, as a remote research intern with the artificial intelligence institute of University of South Carolina (AIISC), I worked with a team on multimodal fact verification. We used Stable Diffusion 2 to enhance the mutlimodal data and built more robust datasets (FACTIFY3M) by injecting adversarial attacks in the form of fake news. This research, which was accepted at EMNLP 2023, gave me firsthand experience in working on multimodal AI.

During my internship at IBM Research, I benchmarked open-source LLMs for enterprise tasks and developed a 3-stage LLM-based approach for NBA (next best agent) recommendation in cold-start settings, work that was accepted at IAAI-24 (co-located with AAAI-24).

My Work at slice

Since joining slice, I've had the opportunity to build several impactful systems.

sliceMCP: I architected the pipeline that combines a Confluence qdrant vector database with a KG to answer complex questions, reducing the onboarding time for new projects from weeks to just a few hours. To make it fast, I designed a data pipeline using Merkle trees that improved synchronization speeds by 100 times for vector database.

Pay-Via-Voice: Built a voice-first payment system using Whisper ASR and multi-agent orchestration. We achieved 1.013s average transaction latency with full UPI integration and policy validation.

ConvoBot: Developed a context-aware chatbot using GPT-4o prompt-chaining that achieves 98.9% precision, 96.7% recall, and 37-42% end-to-end resolution rates.

SMS Processing: Inferenced LLMs like Qwen2.5-3B and Llama-3.1-8B on GPU with batch inference to reduce latency. Got an accuracy of 79% with 0.55s per 3 SMS. Then, trained HRM (Hierarchial Reasoning Model) using DAPT (Domain Adaptive Pre-Training) and SFT for SMS categorization, achieving 86% macro-F1 score with latency around 2ms per SMS.

Research Interests

I am interested in developing AI systems capable in symbolic reasoning and build efficient collaborative systems that works alongside people in meaningful ways. My current focus is on building collaborative, agentic models that can understand goals, share context, and adapt through interaction, ideas that connect closely with Human-Computer Interaction (HCI) and Human-Centered AI (HCAI).

Currently Exploring

• Exploring how reinforcement learning can shape LLM personas for collaborative agent systems, specifically through post-training a Qwen3 MoE model
• Currently digging into this paper, which takes an interesting approach—instead of relying on expert-annotated feedback, it uses real human interactions to develop the model's persona
• Experimenting with the Muon Optimizer, which has been showing better training efficiency than traditional optimizers in recent benchmarks
• On the implementation side, diving into a source code that has claimed to have improved performance over flash-attention to see where I can squeeze out performance gains
• Learning Triton along the way since the performance optimization code is written in it