s
sukhman_sarann

Sukhmandeep S

@sukhman_sarann
5.0(2)

ML Engineer, LoRA fine tuning, RAG systems, model evaluation

India
Hindi, English, Punjabi
About me
I am a machine learning engineer. I fine tune models, and I measure whether the fine tuning actually worked. Image models: LoRA and DreamBooth on Stable Diffusion, using a LoRA implementation I wrote from scratch rather than importing a library. Language models: production RAG with inline citations and guardrails that block answers the sources do not support. Before handover I test on examples you choose and send you the before and after. If it does not beat the baseline, I say so. Python, PyTorch, Hugging Face, FastAPI, Docker. Ask and I will send you the public repositories.... Read more

Skills

s
sukhman_sarann
Sukhmandeep S
Offline • 

See my services

Statistical Modeling & Analytics
I will be your data analyst, visualization expert on python, excel, powerbi
5.0(2)
Product Analytics
I will be your data analyst, excel expert, data cleaning, dashboards

Portfolio

Work experience

NucleoVir_Therapeutics

AI Engineer

NucleoVir Therapeutics • Full-time

Mar 2026 - Present • 8 mos

•Develop and integrate protein language model and structure-prediction workflows for sequence design and AI-driven biomolecule discovery. •Build and maintain automated evaluation, filtering, and ranking pipelines for generated protein candidates by combining structural predictions with sequence- and structure-based criteria. •Design and operate reproducible GPU/HPC workflows for large-scale structure prediction, model inference, and candidate evaluation. •Contribute to AI infrastructure and data-pipeline development supporting biomolecular design workflows in collaboration with computational biology and R&D teams.

National_Institute of Technology Warangal

Machine Learning Engineer

National Institute of Technology Warangal • Freelance

Aug 2025 - Feb 2026 • 6 mos

•Deployed lightweight ML/DL models for fully offline classification and regression on Android, Raspberry Pi, and NVIDIA Jetson Nano devices. •Reduced model size from 10 MB to 3 MB through quantization, achieving 230–350 ms end-to-end inference latency on constrained edge hardware. •Integrated models into a Flutter application with on-device SQLite caching, maintaining <20 MB memory usage and <20% CPU utilization on Android. •Built Dockerized inference pipelines and mentored 10+ interns in ML, deep learning, IoT, edge deployment, and real-time sensor integration.

2 Reviews
5.0

(2)
(0)
(0)
(0)
(0)
Rating Breakdown
  • Seller communication level
    5
  • Quality of delivery
    5
  • Value of delivery
    5
1-2 out of 2 Reviews
Sort By
Most relevant
    B
    image-docs

    bayareasports

    US

    United States

    5

    Sukhman came through when I really needed support last minute! I am working on a massive project and to track all the deliverables, my organization is using an excel spreadsheet. I spent a good 3 days on my own to try to figure out formulas on my own to calculate due dates. Decided to work smart and...

    $100-$200

    Price

    1 day

    Duration

    gig

    Statistical Modeling & Analytics

    Helpful?
    Yes
    No
    I

    instant_pickup

    GB

    United Kingdom

    5

    He is one of the best people here

    $50

    Price

    2 days

    Duration

    gig

    Statistical Modeling & Analytics

    Helpful?
    Yes
    No