next up: technical intern (spring) @ cisco
hi, i'm neuralnets.
i train models &
build systems.
final-year computer science student working across machine learning and systems — rl research, low-resource machine translation, multilingual nlp, and infra at scale.
- papers
- 5
- technical blogs
- 20
- things built
- 36+
- venues
- ICLR · ACL · EACL
INTERSPEECH · COLM
selected papers
all papers-
Scripts Through Time: A Survey of the Evolving Role of Transliteration in NLP
Exploration of transliteration's impact on NLP.
-
Preferences of a Voice-First Nation: Large-Scale Pairwise Evaluation and Preference Analysis for TTS in Indian Languages
Large-scale pairwise evaluation and preference analysis for TTS in Indian languages.
speecharxiv ↗ -
RiddleBench: A New Generative Reasoning Benchmark for LLMs
Benchmark for evaluating LLM reasoning capabilities.
latest writing
all 20 posts- 01 turboquanthow polarquant and quantized johnson-lindenstrauss combine to achieve 3-bit kv-cache compression with zero memory overhead
- 02 how do keyword spotting models work?audio features, neural networks, training, and continuous detection
- 03 how does dynamic memory allocation work?the heap, malloc, free, and allocator internals
- 04 how does interprocess communication work?pipes, message queues, shared memory, and synchronization
things i built
all projectsml papers in code
implementing research papers and deep learning concepts over time, with links to each paper as it is completed.
view on github ↗ cfrom scratch in c
college concepts rebuilt from scratch in c — video extraction, elevator logic, stacks, queues and every flavour of linked list.
view on github ↗ python · numpyneugrad
a lightweight autograd engine mimicking pytorch — tensors, activations, layers, losses, backprop, convolution, adam & sgd.
view on github ↗research question, correction or collaboration?
get in touch