ASH

AI/ML Engineer

Building the intelligent web one model at a time.

I'm Ashmeet Singh Sandhu — a shipping production AI systems and end-to-end web products from idea to scale.

View Services
Resume ready
Ashmeet Singh Sandhu
About

A builder at the edge of AI and the web.

agent.py
from langchain.agents import create_agent
from openai import OpenAI
import pinecone

# Initialize RAG & Agent
index = pinecone.Index("ashmeet")

agent = create_agent(
  model="gpt-4o",
  tools=[search, code_exec, vector_store],
  system="You are Ash — Ashmeet's AI."              
)

agent.run("ship a startup.")

I design and ship intelligent products — from RAG pipelines and ML systems to polished full-stack front-ends. My favorite zone is where elegant UI meets deep AI infrastructure.

Education

Where I studied.

  1. 2024–2028

    B.Tech in Computer Science and Engineering

    IIIT Una. CGPA 8.87. Focusing on AI systems, LLMs, and production ML pipelines.

  2. 2023–2024

    Senior Secondary

    Sri Gururam Rai Public School, Sri Ganganagar.

DSA Submissions

Live problem-solving activity from Codolio across LeetCode, Codeforces & CodeChef.

ashmeet555 codolio
Submissions: —Questions: —Max. Streak: —Current Streak: —

GitHub Contributions

My recent code contributions and open-source activities.

@ASHMEET555
Experience

A short, fast trajectory.

Click QueueBuster or the research internship for full case studies — architecture, challenges, and stack.

May 2026 — Present
QueueBuster India logo

AI Agent Development Intern

QueueBuster India
  • Built the fraud & anomaly detection layer on ~12M live invoices in QueueBuster Helix across merchant, customer, and global risk surfaces.
  • Delivered production-ready rule engines and Go APIs over multi-tenant PoS, payment, and settlement traffic with RBAC-protected scoring paths.
  • Trained ML shadow scoring to ~0.95 AUC and pushed realtime inference to ~200–210 ms with L1 + Redis caching (p95 under 500 ms).
  • Turned detections into analyst-ready investigations and share-gating decisions so fraud signals could drive real product actions.
Open full case study →
Dec 2025 — May 2026
IIT Mandi logo

Research Intern

IIT Mandi
  • Built a deployable multimodal ECG–language pipeline for generalized clinical ECG understanding under tight compute budgets.
  • Replaced a heavy LLM with a compact SLM, ECG-ViT encoder, and modality connector while keeping multi-task reasoning.
  • Shipped full training, inference, and evaluation across report generation, waveform localization, and ECG question answering.
  • Validated compression trade-offs: 68.26 macro-AUC on PTB-XL SUB, ~5.99% ROUGE-L retention loss, plus IoU and QA metrics.
Open full case study →
2024 — Present

AI & Full-Stack Builder

Open Source & Independent
  • Shipped MediNodus, FlashPoint, ChainAuditAI, and triage research platforms end-to-end across product and research surfaces.
  • Built Live RAG systems, local medical VLM pipelines, and Ethereum-backed audit trails for trustworthy AI workflows.
  • Designed FastAPI backends and React product UIs that turn ML prototypes into usable tools for real users.
  • Mentored juniors across ML concepts, backend patterns, and shipping habits so projects moved from notebooks to demos.
 
Stack

Tools I use to build intelligent products.

Python
PyTorch
React
Next.js
TypeScript
FastAPI
Hugging Face
Go
Python
PyTorch
React
Next.js
TypeScript
FastAPI
Hugging Face
Go
Python
PyTorch
React
Next.js
TypeScript
FastAPI
Hugging Face
Go
Docker
AWS
PostgreSQL
MongoDB
Redis
Kafka
Git
Linux
Docker
AWS
PostgreSQL
MongoDB
Redis
Kafka
Git
Linux
Docker
AWS
PostgreSQL
MongoDB
Redis
Kafka
Git
Linux

Languages

PythonC++TypeScriptJavaScriptSQLGoHTML / CSSBash

AI & Machine Learning

PyTorchDeep LearningLLMs & RAGLangChainLangGraphComputer VisionECG-ViTSLM CompressionModel Fine-tuningHugging FaceEmbeddingsSHAP / LIMEXGBoostLightGBMIsolation ForestAutoencoder

Frameworks & Libraries

FastAPIFlaskNext.jsReactTanStack QueryStreamlitPandasNumPyScikit-learnspaCy

Data & Backend

PostgreSQLClickHouseMongoDBRedisQdrantKafkaREST APIsGin (Go)RBAC / AuthZ

Tools & Platforms

DockerAWSGitCI / CDGitHub ActionsLinuxExperiment TrackingSystem Design
Selected work

Products, models and platforms I've shipped.

Click any project to open a detailed case window — demos, stack, and walkthrough.

Explore more projects
Credentials

Validated expertise in systems and analytics.

Achievements

CodeChef 4-Star · Codeforces Specialist (1530)

Issued Ongoing

1000+ problems solved across major platforms with consistent competitive programming performance.

Highlight
ICPC

AIR 720 · ICPC

Issued 2025

National-level competitive programming placement demonstrating algorithmic problem-solving under pressure.

Highlight
TechOlympics

Global Rank 11 (India Rank 1)

Issued 2025

Top national finish in TechOlympics 2025 with strong systems and AI problem-solving outcomes.

Highlight
Community

4+ Hackathons

Issued 2024–2026

Built and shipped AI prototypes under time pressure across product, research, and systems tracks.

Highlight
AlgoUniversity

Graph Theory Programming Camp

Issued Certificate

Participated under mentorship of a Codeforces Master and completed 17 advanced graph problems.

View certificate
BIT Mesra

IEEE Mega Project 8.0 — Participation

Issued Hackathon

Certificate of participation for IEEE Mega Project 8.0 organized by BIT Mesra, Ranchi.

View certificate
IIT Kharagpur

IIT Kharagpur Hackathon

Issued Hackathon

Hackathon participation certificate from IIT Kharagpur.

View certificate
Meta

Meta Hacker Cup

Issued Contest

Participation certificate from Meta Hacker Cup competitive programming contest.

View certificate
Infosys Springboard

Infosys Springboard Certificates

Issued 15 courses

Completed Infosys Springboard modules across AI, GenAI, deep learning, NLP, computer vision, data science, and more.

View all certificates
Contact

Let's build something intelligent together.

Whether it's an AI feature, an MVP, or a full product — I'm open to internships, founding-engineer roles, and collaboration.

Una, Himachal Pradesh• (GMT+5:30)
View on Google Maps