Austin Dcosta

ai research → infrastructure.

building things, looking for internships

bengaluru, india (ist)

ai can write code. ai can't have taste.

ludwig van beethoven — joseph karl stieler, 1820

// about

i liked taking things apart as a kid. remotes, toys, anything that didn't get taken away fast enough. computers were the first thing i could take apart and put back together without getting in trouble.

i study ai/ml at pes university in bengaluru. i started in research and co-authored a paper on discovering llm prompting techniques with knowledge graphs, which was accepted at icann 2026. somewhere along the way i got more curious about what sits underneath: queues, storage engines, processes that crash and come back. that's mostly what i build now.

i also ta data structures & algorithms, do competitive programming, and take quizzing more seriously than i should.

codeforces specialist (1580)
430+ leetcode problems
paper at icann 2026 (padua, italy)
b.tech ai/ml, pes university, 2023–present

// work

jun – aug 2026 · bengaluru
summer analyst intern, wealth management technology — morgan stanley
built an ai-powered validator that checks team names against compliance policy on its own, saving legal and compliance officers weeks of manual review. optimized 10+ sql queries for internal reporting, and shipped backend changes to the team's java codebase.
learned: how to work properly inside a big collaborative team.
- java
- sql
- llms
aug 2025 – present
teaching assistant, data structures & algorithms — pes university
run labs and debugging sessions on trees, graphs, recursion and complexity, and write problem sets.
feb – apr 2025
project intern, advanced driver assistance — centre for innovation & entrepreneurship, pes university
real-time vehicle and pedestrian detection with yolov8 (85% accuracy). cut inference latency 30% by switching models based on system load. lane detection and multi-object tracking on embedded hardware.
learned: how vision models actually behave outside a notebook.
- pytorch
- opencv
- yolov8
- embedded

// projects

the bottleneck was fsync, not the network

active2026

distributed job queue

a durable, sidekiq-style job queue written in c++.

solo

c++
sqlite
posix sockets
docker

read the case study →

highlights

at-least-once delivery with lease-based claiming
concurrency safety handed to sqlite's file locking: zero double-claims across 32 threads and 5,000+ jobs
found that per-commit fsync, not networking or threads, capped throughput: ~150 jobs/s fully durable vs ~1,700–2,800 relaxed (11–18x)
fixed a query-planner regression with explain query plan
fixed a stale-lease race with a fencing token
sigkill mid-load across 10+ restarts with zero acknowledged jobs lost

finished2026

fault-tolerant ml training runtime

a multi-process ml training runtime that survives worker crashes.

solo

c++
posix (fork/exec/waitpid)
linux

highlights

parallel workers across cores that keep training when one dies
mmap'd dataset loading
crash-safe checkpointing for fast recovery

published2026

knowledge-graph-guided prompt discovery

finding new llm prompting techniques by treating it as link prediction on a knowledge graph.

co-author (research) · with abhay patil, dr. bhaskarjyoti das (mentor)

python
knowledge graphs
llms

paper

highlights

built a heterogeneous graph of prompting techniques, use cases and their components
discovered compositions beat baselines by up to 27% on truthfulqa-mc1 and 18% on tweeteval-sentiment
accepted at icann 2026, padua (poster)

also

bustub2026
in progress: buffer pool + b+ tree done
building pieces of a relational database engine from scratch (cmu 15-445).
- c++
highlights (2)
- buffer pool manager with page replacement and disk i/o
- b+ tree index with ordered lookups and range scans
politicalsim2026
finished
a simulated democracy where ai politicians learn to win elections, and voters push back.
- python
- mesa
- gymnasium
- stable-baselines3
- networkx
- tensorboard
highlights (4)
- ai candidates trained with reinforcement learning (ppo) shift positions across multiple policy issues to win votes
- voters live in a small-world social network and influence one another's views
- shocks like economic crises, pandemics and wars hit the electorate mid-run
- elections every 20 steps; tracks polarization and voter satisfaction over time