Ankit Dahal

About

I took the decision to leave my role in 2025 to create a path for myself to pivot into AI/ML engineering. Years of foundational AI knowledge have been compressed into months of strenuous study and application through personal projects. This has allowed me to move exponentially faster than I would have been able to otherwise, building a strong understanding of modern AI from first principles.

I am looking for technically challenging problems that matter. I strive to be a strong generalist, diving deep when needed, and am passionate about a wide range of topics from AI, scalability, cybersecurity, to biology, psychology, and cosmology.

Current focus

My primary focus is landing the right role. Projects keep me sharp, productive, and building while I search.

Current (Primary) Adversarial capabilities research across major social media platforms with computer-use agents and reinforcement learning.
Future interest Fully autonomous AI agent swarms for task specific web scraping. Decomposing tasks and orchestrating sub-agent hand-offs with coordination & self-triage. Autonomous discovery and use of publicly facing APIs (not just webpages).

Projects

Adversarial Agents Research

In progress Current AI Safety

Mapping AI-enabled manipulation across major social media platforms. Studying coordinated inauthentic behavior at scale.

View details

Mapping the threat landscape of AI-enabled manipulation across major social media platforms. Studying coordinated inauthentic behavior: how bot-driven narratives (astroturfing) exploit platform mechanics and cognitive biases at scale.

5+ Platforms
AI + Psychology + Cybersecurity Interdisciplinary

Technology: Qwen3.5-35B-A3B (student), Qwen3.5-397B-A17B (teacher), Reinforcement learning (web-browser gym environment), python

Adversarial AI Coordinated agents Platform analysis Cognitive psychology AI safety
Project Outline →

Large Language Model Pretraining

MVP Pretraining

416M parameter transformer inspired by the gpt and llama papers, pretrained on 10B tokens using 8xA100 GPUs.

View details

End-to-end pretraining of a custom, self-built 416M parameter GPT/Llama-style transformer on FineWeb-Edu (10B tokens). Trained on 8xA100 GPUs using Distributed Data Parallel with custom training loops and memory optimizations.

416M Parameters
10B Tokens
8x A100 GPUs
PyTorch, DDP Custom training loop Chinchilla scaling law RoPE, RMSNorm, SwiGLU, GQA/MHA, KV cache Flash attention
Code → Model Checkpoint →

Legal RAG System

MVP RAG / Agents

Production-grade agentic RAG for legal documents with hybrid dense + sparse retrieval.

View details

Production-grade agentic RAG system for legal documents. Full pipeline from data sourcing through retrieval, reranking, and agent orchestration. Hybrid search combining dense embeddings (bge-m3) with sparse retrieval (BM25).

bge-m3 ChromaDB Elasticsearch bge-reranker Gemini-2.5 Docker GKE

Embeddings: bge-m3, gemini-embedding-001

Indexing: ChromaDB (HNSW), Elasticsearch (BM25)

Retrieval: Hybrid search with RRF, convex combination

Agents: Conversational + search agents, planning, self-triage

Training: 560M parameter model fine-tuning

Code → Training Infrastructure →

Llama 3.1 8B Instruction Tuning

Fine-tuning

LoRA fine-tuning with +52% IFEval improvement using ~$10 compute.

Details

LoRA fine-tuning of Llama 3.1 8B for instruction following. Achieved +52% IFEval improvement (200 → 305/834) with ~$10 of compute across 4-bit, 8-bit, and BF16 quantization experiments.

LoRA Unsloth Quantization IFEval W&B
Code →

Transformer from Scratch

Fundamentals

Clean PyTorch implementation with modern LLM improvements.

Details

Clean PyTorch implementation of "Attention Is All You Need," extended with modern LLM architectural improvements: RoPE, SwiGLU, RMSNorm, GQA.

PyTorch MHA SwiGLU/ReLU RMSNorm
Code →

LLM VRAM Calculator

Tooling

GPU memory estimator for LLM training and inference.

Details

GPU memory estimator for LLM training and inference. Supports dense and MoE architectures with breakdowns of weights, gradients, optimizer states, activations, and KV cache.

Gradio HuggingFace MoE Mixed Precision
Live Demo → Code →

Global Placement AI Initiative

Educational AI

Support for international students' global placements.

Details

AI-first support system to help international students improve application quality, interview readiness, and global placement outcomes. Prioritizing outcomes over profit, severely undercutting traditional methods. I built the core technology, currently hands off on this project as it has been handed over to others. Project demo available on request.

Social impact International Students AI for good Economic opportunity Autonomous processing

Writing

Technical Writeup

Building a Production-Grade Legal RAG System

End-to-end design notes covering hybrid retrieval architecture, embedding strategies, reranker integration, agentic orchestration, and lessons learned building a RAG pipeline for legal documents.

Open to Opportunities

Looking to join teams building interesting AI systems.

View Resume GitHub LinkedIn