B.Tech, Computer Science
Jawaharlal Nehru Technological University, Hyderabad
- computer science undergrad in hyderabad.
- the aquaponics project ran alongside it from 2022 to 2024.
PLAYER 1 · AI ENGINEER · LAS VEGAS
Sai Anirudh Siddi
hi, i'm ani. i work with ai for a living. after hours i build hobby projects for fun and solve problems with technology.
master of engineering in cs from the university of cincinnati. now in las vegas, fuelled by a desire to do something different. something great.
About Sai Anirudh Siddi
ani@vegas ~ $ ani --about
ani@saianirudh.blog
- class
- ai engineer, product dev
- base
- las vegas, nv
- status
- ai fellow at handshake aivolunteer engineer at an ngobuilding toki (in progress)
ani@vegas ~ $ ani --abilities
- ai engineeringllm apps and agents, prototype to production
- product devidea to shipped product, end to end
- data + ml pipelinespipelines that stay up at 3am
- iot + mlsensors in, decisions out
- interpretabilityopening models up to see what they compute
Quest log · 6 entries
Writing
Older entriesSee all 6 posts →
Most recent · 4 completed · 1 in progress
Projects
- MECH INTERP · FEATURED
GlassBox
x-ray for medical llms. 15+ sae features live during inference, flags when the model is unsure of itself.
Read the series · 3 posts → - IN PROGRESS
Toki
fully offline meeting intelligence: asr, diarization and a 4-bit qwen3-4b on a 4gb gpu. live transcription from 21s to ~1s, 9.2% wer. nothing leaves the laptop.
In progress - NASA · AI
ExoSeeker
exoplanets in kepler data, >90% accuracy.
LOOTbest use of nasa data, space appsRepo ↗ · 1 post (opens GitHub in a new tab) - MEDTECH · IOT
NeedleHelp
iot + ml for real-time robotic control.
LOOT1st of 140 teams, ohio's largest hackathonRepo ↗ (opens GitHub in a new tab) - IOT · ML
Aquaponics
taught a working fish farm to run itself. 1,000 readings a day, 40% less babysitting, 20% more yield.
Repo ↗ (opens GitHub in a new tab)
Trophy room
Wins and press
- WIN2025Best Use of NASA Data, NASA Space Apps ChallengeExoSeeker · 2,000+ participants(opens in a new tab)
- WIN20251st Place, Best Project Overall, Ohio's largest hackathonNeedleHelp · 140+ teams · via 1819 Innovation Hub(opens in a new tab)
- PRESSNov 2025Cincinnati teams blast off to the global stage in NASA hackathonuc.edu(opens in a new tab)
XP log · drag to scrub through the run
Full page →Experience and education
Automating Aquaponics
Startup incubator project, with professors and a local business
- taught a working fish farm to run itself: iot monitoring across water, ph and environment sensors, 1,000+ readings a day, 40% less manual babysitting.
- ml yield forecasting that contributed to a 20% improvement in crop yield.
- tableau dashboards that drove a 10% profit increase for the business; research paper under peer review.
Data Engineer
Dsign Code LLC
- sql and pl/sql reporting workflows across 20+ enterprise modules.
- analyzed 50k+ monthly transactional records; improved system performance by 15%.
Software Engineer, Digital Forensics
DRDO, Defense Research and Development of India
- processed 700+ structured and unstructured forensic artifacts per run.
- built data ingestion, management and validation pipelines for downstream investigative analysis.
- improved system reliability and investigative efficiency by 30%.
Master of Engineering, Computer Science
University of Cincinnati
- grad school in ohio.
- best use of nasa data at nasa space apps (exoseeker), 1st of 140+ teams at ohio's largest hackathon (needlehelp).
Data Engineer, Predictive Analytics
Great American Insurance Group
- analyzed 1m+ underwriting and litigation records; 10% improvement in decision accuracy.
- built optimized data management pipelines and reporting across multiple systems.
- validated an ai-assisted analytics tool with stakeholders; 60% better processing efficiency.
AI Fellow
Handshake AI
- author adversarial evaluation sets and rl environments for frontier llms, with rubric-based reward functions; over 50% stumping rate on multi-step quantitative reasoning.
- build and curate labeled evaluation datasets, refining prompts against model reasoning traces.
- score outputs against calibrated rubrics that serve as rl reward signals.
ExpandRange
in the works
- building our own ai products and services.
ANIRUDH
want to chat? reach out to me here.
i'm always looking to connect and meet new people. hmu.