# Harsh PV

> Founding Engineer @ Anchr | a16z Speedrun | CS @ UIUC

Location: San Francisco Bay Area, United States
Profile: https://flows.cv/harshpv

Software Engineer | CS @ UIUC | Ex-Cisco | AI + Full-Stack + Research | 7x Hackathon Winner

I'm a Master's in Computer Science student at UIUC, passionate about building high-impact software at the intersection of AI, systems, and product engineering. I've worked on fast-paced engineering teams at Cisco and Novartis, contributed to NSF-funded research projects with UIUC and USC, and built multiple AI-powered tools that span full-stack development, LLM workflows, and VLM optimization.

🛠 Technical Strengths:
Python, C++, Java, JavaScript, Go, Django, Flask, React, Node.js, PyTorch, MongoDB, PostgreSQL, Kafka, Redis, Docker, Terraform, AWS, GCP, LangGraph, Triton, DALI

🧠 Areas of Interest:
LLMs, multi-agent workflows, retrieval-augmented generation (RAG), GPU inference optimization, knowledge graphs, scalable backend systems, and dev tooling

🚀 Notable Work:

Cisco: Built config translation engines (+30% adoption), automated network deployments (–60% time)

UIUC Research: Reduced inference time from 6 hours to 2 mins using NVIDIA Triton & RAPIDS; fine-tuned VLMs for structural defect detection

SALT Lab: Tackled hallucination in LLMs via knowledge graph-based RAG systems

Hackathons: 7-time winner including NCSA Ashby Hackathon, SIH 2020, and McAfee Hackathon

I thrive in cross-functional teams, rapidly iterate, and ship impactful software. Actively exploring full-time SWE roles starting May 2025.

🔗 Portfolio: https://harshpv.vercel.app
💻 GitHub: https://github.com/harshpv07
📫 Let’s connect — open to roles in AI infrastructure, LLMs, distributed systems, or full-stack/backend engineering.

## Work Experience
### Founding Engineer @ anchr
Jan 2025 – Present | San Francisco Bay Area
- Founding engineer building AI agents for the food distribution industry, focused on automating real customer workflows such as order intake, pricing validation, inventory checks, and ERP updates.
- Improved end-to-end agent reliability and task accuracy by ~70% by refining multi-agent coordination, decision logic, and LLM integration, directly impacting customer-facing outcomes.
- Designed MCP-based integrations for core data systems including Supabase and NetSuite, enabling agents to operate on live, production data with low latency and high correctness.
- Built a centralized tool registry to streamline how agents discover and use tools, reducing token usage by ~40% while improving response speed and cost efficiency.
- Implemented episodic and long-term memory to help agents retain context across interactions, resulting in more consistent, personalized, and actionable responses.
- Shipped fine-grained access controls and authentication for agents and tools, ensuring secure, tenant-aware execution while maintaining a smooth user and operator experience.

### Software Engineer @ University of Illinois Urbana-Champaign
Jan 2024 – Present | Champaign, Illinois, United States
- I worked on enhancing GPU accelerated Inference times using frameworks like Nvidia Triton, DALI & RAPIDS. By implementing optimization techniques such as Quantization and Batching, I was able to significantly improve system performance and efficiency. Collaborating with cross-functional teams, I achieved remarkable reductions in latency and overall system performance, contributing to the success of the AIOps initiatives.
- Worked on Multi modal Biomedical Image registration using ECC and FFT Image transformations.

### Teaching Assistant @ University of Illinois Urbana-Champaign
Jan 2024 – Jan 2024 | Champaign County, Illinois, United States
TA for CS 517 - Scientific Visualization

### Software Engineer @ University of Illinois Urbana-Champaign
Jan 2024 – Jan 2024 | Champaign, Illinois, United States
- Collaborating with Dr. Huang at SALT Lab to tackle hallucination and factuality loss in LLMs using knowledge graphs-based RAG systems.
- Analyzed conversational emotions within victim-dispatcher interactions for the VicSim project, aiming to glean insights into emotional dynamics.
- Finetuned LLama - 7B and NALLM using SFT to generate Neo4J Ciphers for Knowledge graph reconstruction.
- Proficient in GPT4, Llama 7B, RAGs, Langchain, Llamaindex, Langgraph, KGs, Pinecone, FAISS, etc

### Software Engineer @ Cisco
Jan 2021 – Jan 2024 | Bengaluru, Karnataka, India
• Developed new features for Cisco’s configuration translation and configuration migration engine using ReactJS, and Django which resulted in a 30% increase in the adoption of Cisco products.
• Engaged in client interactions to design, deploy, and troubleshoot networks, while demonstrating expertise in configuring and managing network appliances such as Cisco switches, routers, firewalls, and cloud security solutions.
• Automated network deployments for global clients using frameworks such as Terraform & Ansible which reduced network deployment times by 70%.
• Obtained certifications such as CCNA, CCNP R&S, CCNP SCOR, DevNet, Meraki ECMS & AWS CSA.

### Software Engineer- Backend @ Novartis
Jan 2021 – Jan 2021 | Hyderabad, Telangana, India
Roles:
• Developed an analytics dashboard using React, ChartJS, and Django, while creating ETL pipelines from various ITSM systems like Salesforce and ServiceNow through Snowflake.
• Developed an ML-powered capacity planning service using boosting algorithms such as XGBoost to benchmark 25+ project and team KPIs. This service was later integrated into the dashboard.
• Deployed the entire project on Azure cloud and gained expertise in Azure services such as Blob storage, AKS cluster, Azure VMs, etc

Tech stack:
-> Dash, Plotly, React JS, Steamlit, Pandas, Flask, Rest APIs, Snowflake & Azure cloud.

### Research Assistant @ Vellore Institute of Technology
Jan 2019 – Jan 2021 | Greater Chennai Area
Roles:
-> Research done in motion detection using background and foreground subtraction.
-> Assisted in development of rotational invariant Image captioning system by integrating RotNET with Image captioning.
-> Looked into the feasibility of rotational invariant CNN for Image captioning system.
-> Assisted in the development of Image captioning software for surveillance applications using deep learning techniques like CNN, LSTM, etc.
-> Helped to improve the accuracy of the model by using different image augmentation techniques for a CCTV application.

### Software Engineer @ Siam Computing
Jan 2019 – Jan 2019 | Greater Chennai Area
Roles:
-> One of the first developers to work on the brand new RASA X which is a chatbot based Assistant. Responsible for building and modelling a smart assistant using RASA X for the organization.
-> Development of a prototype of a smart attendance system for a popular school which uses the facial encodings to recognize faces and map them with the name. Deployed the entire prototype locally with a Django Backend and MongoDB to monitor Attendance logs and Teacher | Student Status.
-> Development of a smart chatbot assistant on Dialogflow for a client of the organization which consists of multiple dialog routes for various functions like User Handling, Recruitment, Sales, FAQs. Successfully integrated and deployed on a PHP Backend.


## Education
### Master's degree in Computer Science
University of Illinois Urbana-Champaign

### Bachelor's degree in Computer Science
Vellore Institute of Technology


## Contact & Social
- LinkedIn: https://linkedin.com/in/harsh-pv-988b66146
- Website: https://harshpv.vercel.app/

---
Source: https://flows.cv/harshpv
JSON Resume: https://flows.cv/harshpv/resume.json
Last updated: 2026-04-05