Ash Srivastava

Aishwarya 'Ash' Srivastava

CS @ ASU · SWE @ ASU DT · ML @ ARC LAB

About me

I like working on ambitious problems around performant AI infrastructure and distributed systems. I also love studying reasoning representations in large models, especially involving latent computations, to design more safer and reliable AI systems.

Currently

  • designing latent methods for faithful agents
  • building more in distributed inference systems
  • teaching others about tech

focus

  • systems, infra, full stack
  • reasoning reliablity & safety in agents
  • ai labs, space, markets

Roadmap

product studio inc.

2026

building user personalized layer for agents

arc lab

2026

working on latent space reasoning in llm architectures

piramid

2026

building inference system for rag applications; inspired from deepmind RETRO

nexhacks

2026

first east coast winter and CMU visit; built an agentic scraper and shipped it on site

the sudden visit

2025

long layover in Boston, wandered MIT; huge motivation jolt that pushed me harder

calhacks

2025

first SF hack; built an agentic workflow analytics tool and came home with a bigger fire

kaelum

2025

implemented RL + LLM tree ideas from papers; learned how far careful engineering goes

ais workshops

2025

taught and wrote the LLM/RLHF-from-scratch notebooks; best way I learned production AI

zoomhacks

2025

won 1st place; built 2d game w unity

google hackathon

2025

won 2nd place; built rag based finetuning ml pipeline

godfather & ai pedia

2025

built infra-heavy projects: custom images, GPU clusters, and an org wiki; realized I like AI + infra

rise lab

2025

joined rise lab HCI group; building transformer/vae architectures for trajectory prediction with human modeling

soda platform

2025

ran internal infra/platform work—bots, docs, deploys—which pushed me into deeper SWE and devtools

devhacks hackathon

2025

won 1st place; built social engineering ai voice agent

decision theater

2025

part-time SWE at an ASU research unit, building tools for data analysts

SoDA

2025

joined software developers association focusing on tech

intel hackathon

2025

won with a lidar anomaly project; first time shipping AI with strict hardware limits

devhacks'24

2024

built agentic fashion recommendation system; got more deep into vision agents

sportsitters

2025

ground-up freelance build; long hours shipping for a client taught real delivery pressure

scalePU hackathon

2025

shipped real time multimodal agents for zoom tutoring sessions

sparkyai & ghost lab

2024

built agentic copilots and tinkered with robots; decided AI+systems is the path

interest in LLMs

2024

turning point: daily LLM papers and experiments made AI the career path

HackSoDA

2024

shipped agentic scraper for student college assistance

sunhacks'24

2024

shipped TUI coding agent; got into multi-task planning agents

asu move

2024

new country, found footing, joined AIS and started chasing AI papers daily

job hunt lull

2024

sent a ton of on-campus apps, got crickets; doubled down on projects and papers instead

codeday

2024

won a national hack; learned to navigate messy orgs while still shipping

ctrl hacks

2023

won an international hack with a student copilot; first real taste of team shipping

happyhacks

2023

won another international hack, built a 24/7 live video chat platform

CSYA & Hack United

2023

built Discord, Notion, and GCP integrations plus headless CMS sites for student orgs; dug deep into tooling

freelance & mobile

2023

built apps for a friend’s company; learned to ship fast for real users

ownboon

2022

first startup attempt; ran community, ops, and product—learned by doing everything

school friction

2022

didn’t fit the local grind; coding after class taught me to self-direct

techxcelerate

2022

won my first international hack; realized I liked building in public

full stack grind

2021

daily MERN practice, shipping small sites, and getting hooked on the craft

game addiction break

2021

dropped games after a family crisis; swapped hours into code and routines

pandemic pivot

2020

first laptop for classes; fell into code instead of games and never really left

curious kid

childhood

quiet, often lost, but always trying to figure out how things worked