press any key to continue 

Rishi Shiramshetti

SCU B.S. CS '27 // ML Systems & Autonomous Agents

status: online uptime 00:00:00 · hover any text to corrupt it · type in the shell

i am a computer science student at santa clara university (b.s. '27) focusing on machine learning systems, autonomous agents, and high-performance computing.

currently, i am a campus strategist at perplexity, driving referral campaigns and AI-search workflows. recently, my team won 2nd place at the snowflake hackathon for toolforge and an honorable mention at the openai x modal hackathon for rustforge.

i am passionate about building agentic frameworks, fine-tuning LLMs, self-evolving systems, and optimizing GPU inference pipelines.

> experience

[02]

perplexity

campus strategist : santa clara, ca

  • drove perplexity and comet campus signups via referral campaigns and peer-to-peer demos.
  • generated $5,000+ in referral revenue through campus marketing campaigns and AI-search activations for research and coding workflows.
  • partnered with ACM at SCU and AI Collaborate (AIC); collaborated with perplexity's growth team to improve adoption.

canyon view technology

technology intern : remote

  • developed a lua script using the haversine equation to improve GPS-based rover navigation precision.
  • conducted comprehensive testing and created technical documentation detailing methodologies and solutions.

> featured projects

[03]

toolforge: self-extending agents

2nd place @ snowflake

langchain / crewai : composio : snowflake : openai embeddings

built a self-extending agent system that generates, validates, and persists reusable tools. features a 5-agent CrewAI pipeline, snowflake vector registry with 1536-D embeddings, and composio remote workbench sandboxing with AST import filtering.

rustforge: self-evolving rust

honorable mention @ openai x modal

python : rust : langchain : modal : pyo3 : hypothesis

engineered an autoresearch system spawning 50 parallel LLM agents per round to rewrite slow python functions into rust via PyO3 across modal serverless containers. achieved a 446–548x speedup on a 1M-element baseline with property fuzzing verification.

[benchmark_visualizer] 548x peak
python (1M) [░░░░░░░░░░░░░░░░░░░░] 1.0x · 1820ms
rust (pyo3) [░░░░░░░░░░░░░░░░░░░░] 548.2x · 3.3ms

cineflow: intelligent video transitions

1st place @ adobe

react : tensorflow : opencv : aws lambda + efs

fine-tuned a vision transformer on 10k clips via tensorflow (23% higher F1 than C3D) to detect optimal cut points. deployed as a GPU-backed microservice on AWS Lambda delivering sub-250ms inference on 1080p footage.

> technical skills

[04]
> languages [core]

Python, C++, Java, JavaScript (ES6+), HTML5, CSS3, Lua

[░░░░░░░░░░░░░░░░░░░░] 90%
> machine learning [ai/ml]

NLP, Computer Vision, Transformers, LLM Finetuning, RL, RAG, Vector DBs

[░░░░░░░░░░░░░░░░░░░░] 95%
> frameworks & libs [tools]

PyTorch, TensorFlow, OpenCV, NumPy, Pandas, FastAPI, React

[░░░░░░░░░░░░░░░░░░░░] 85%
> infra & platforms [cloud]

AWS, Modal, Unsloth, Hugging Face, Docker, Agentic Frameworks, vLLM

[░░░░░░░░░░░░░░░░░░░░] 90%

> awards & honors

[05]
  • h.m. honorable mention — autoresearch systems hackathon (openai, modal, raindrop & antler)
  • best best use of tencent cloud — SCU hack-a-stack
  • 2nd 2nd place winner — cerebral valley x llama lounge hackathon @ snowflake
  • 1st 1st place winner — AIC x adobe express hackathon @ santa clara university
  • schol bronco scholarship — merit-based academic scholarship at santa clara university
rishi_shell — bash
tab completes · ↑↓ history ·
rishi@scu:~$