Ashmit M.-AI Training and LLM Evaluation Expert | Mathematics and Software Engineering
Check rate
Experience
AI Training Expert
Handshake AI
- LLM evaluation, benchmark design, red teaming
- Design Terminal-Bench style agentic tasks (Docker, Python, pytest verifiers) that frontier coding agents fail; multiple tasks accepted with 0/5 pass@5 against the gating model.
- Build reference solutions, synthetic data generators, anti-cheat verifiers, and oracle/no-op validation so every task is fair, deterministic, and solvable by an expert.
- Author context-learning math tasks and multi-hop research prompts; analyze per-sample model outputs to find recurring failure modes and harden tasks against them.
Software Engineer II
Infosys Ltd.
- Built a microservices pipeline processing 2M+ daily events; cut end-to-end latency 40% with query tuning and Redis caching.
- Built Bayesian models for customer lifetime value prediction; improved accuracy 23% and guided Rs 4.2 Cr in budget decisions.
- Migrated a Java monolith to Kubernetes on AWS, delivering 99.97% uptime with 25% lower infrastructure cost.
Mathematics Specialist
Turing
- RLHF, math reasoning evaluation
- Create original competition-level math problems in Jupyter Notebook to stress test GPT and Claude models on logic, calculation, and multi-step reasoning.
- Review model responses for faulty logic, hidden assumptions, and calculation errors; write corrected step-by-step solutions verified with Python and SymPy.
- Rank responses and write preference feedback (RLHF) against rubrics; rated among top specialists for accuracy and clarity and helped set the team quality bar.
Software Engineer
Infosys Ltd.
- Shipped Spring Boot and Node.js REST APIs serving 100K requests per day at under 50 ms p99 with 92% test coverage.
- Built an LSTM anomaly detector on system telemetry, cutting detection time from 15 minutes to under 2 minutes.
- Automated CI/CD with Jenkins and GitHub Actions, cutting release time 60% with zero manual errors.
Adversarial ML Defense System
IIT Delhi
- PyTorch, scikit-learn, SHAP, Docker
- Ensemble defense for IoT traffic reaching 96.8% detection on CICIDS 2017 with under 0.3% false alarms.
Mini OS Kernel (xv6)
- C, x86 Assembly, QEMU, GDB
- Extended xv6 with an MLFQ scheduler, copy-on-write fork, and demand paging; cut context switch time 35%.
Research Intern
Microsoft Research India
- NLP, machine translation
- Improved low-resource machine translation by 4.2 BLEU over mBART using attention models trained on 1.2M sentence pairs.
- Co-authored a paper submitted to ACL 2024.
Software Engineering Intern
Samsung R&D Institute
- On-device machine learning
- Shrunk an on-device image model 42% with quantization while keeping 98.1% accuracy.
- Shipped TensorFlow Lite object detection running at 30 FPS on mid-range Android phones.
Math Reasoning Evaluation Suite for LLMs
Turing
- Python, Jupyter, SymPy, JSONL
- Built 250+ original math tasks with verified answers; benchmarked GPT and Claude, tagged failure patterns, and produced ranked preference pairs in JSONL for RLHF and DPO training.
Real-Time Code Collaboration Editor
Self
- TypeScript, WebSocket, Yjs, Monaco
- Collaborative editor built on CRDTs supporting 50+ concurrent users with syntax highlighting for 20+ languages.
Industry experience
See where this freelancer has spent most of their professional time.
Experienced in Information Technology.
Business area experience
See which departments and functions this freelancer has contributed to most.
Experienced in Information Technology, Research and Development, Product Development, and Quality Assurance.
Summary
Author original expert-level math and coding problems that frontier models (GPT, Claude, Gemini) fail, write verified step-by-step solutions and grading rubrics, and build Dockerized agentic coding tasks with automated verifiers.
Skills
- Ai Training And Evaluation: Llm Evaluation, Rlhf, Sft Data Creation, Preference Ranking, Reward Modeling Data, Rubric Design, Red Teaming, Adversarial Prompt Design, Benchmark Design, Agentic Coding Evaluation, Error Analysis, Chain-Of-Thought Review, Prompt Engineering, Data Annotation, Quality Assurance
- Mathematics: Probability, Statistics, Linear Algebra, Calculus, Combinatorics, Number Theory, Bayesian Inference, Causal Inference, Optimization, Proof Writing, Latex
- Programming: Python, Sql, Java, C, C++, Typescript, Go, Bash
- Ml And Data: Pytorch, Tensorflow, Scikit-Learn, Hugging Face Transformers, Numpy, Pandas, Scipy, Sympy, Jupyter
- Infrastructure: Docker, Kubernetes, Aws, Git, Github Actions, Linux, Postgresql, Redis, Pytest
Languages
Education
Indian Institute of Technology Delhi
B.Tech · Computer Science and Engineering · New Delhi, India · 8.74/10
Statistics
Experience
Global experience
Expertise
Qualifications
Frequently asked questions
Have questions? Find more information here.
Ashmit is based in Vadodara, India and prefers 100% remote projects.
Ashmit speaks the following languages: Hindi (Native), English (Advanced).
Ashmit has at least 5 years of experience. During this time, Ashmit has worked in at least 10 different roles and for 8 different companies. The average length of individual experience is 1 year and 6 months. Note that Ashmit may not have shared all experience and actually has more experience.
Based on recent experience, Ashmit would be well-suited for roles such as: AI Training Expert, Software Engineer II, Mathematics Specialist.
Ashmit's most recent position is AI Training Expert at Handshake AI.
In recent years, Ashmit has worked for Handshake AI, Infosys Ltd., Turing, IIT Delhi, and Google.
Ashmit is most experienced in industries like Information Technology.
Ashmit is most experienced in business areas like Information Technology, Research and Development, and Product Development. Ashmit also has some experience in Quality Assurance.
Ashmit holds a Bachelor in Computer Science and Engineering from Indian Institute of Technology Delhi.
Ashmit is immediately available full-time for suitable projects.
Daily rate distribution
The rates shown represent the typical market range for freelancers in this position based on recent contracts on our platform.
Average rates for similar positions
Rates are based on recent contracts and do not include FRATCH margin.
The average daily rate is the mean of all daily rates from recent contracts of comparable freelancers on our platform.
The median daily rate is the middle value of all daily rates — half of comparable freelancers charge less, half charge more. Unlike the average, it is barely affected by outliers.
Calculated based on our freelancers’ daily rates as of 29 Oct 2026. Actual rates may vary depending on seniority level, experience, skill specialization, project complexity, and engagement length.
Similar freelancers
Discover other experts with similar qualifications and experience
Abhishek N.
Kiran kumar K.
Ekaansh K.
Chintan P.
Srijan M.
Noushiq M.
Kamal R.
Ajay C.
Shamaila M.
Shekhar K.
Experts recently working on similar projects
Freelancers with hands-on experience in comparable project as a AI Training Expert
Majid K.
Nikhil G.
Stefan O.
David M.
Justas F.
Eric Y.
Robin L.
Benjamin W.
Heena P.
Alessandro P.
Nearby freelancers
Professionals working in or nearby Vadodara, India
Dinmukhabet A.
Khanim A.
Hikmat S.
Javid K.
Sergey F.
Mansur Z.
Emin M.
Elena G.
Beka C.
Gvantsa K.
