AI Data Scientist

ducation : Bachelor’s or Master’s Degree in Computer Science, Engineering, Maths or Science P

erformed any modern NLP/LLM courses/open competitions is also welcomed.
T

echnical Requirements :
S

oft Skills : S

• trong communication skills and do excellent teamwork through Git/slack/email/call with multiple team members across geographies
  G

enAI Skills : E

• xperience in LLM models like PaLM, GPT4, Mistral (open-source models), W
• ork through the complete lifecycle of Gen AI model development, from training and testing to deployment and performance monitoring. D
• eveloping and maintaining AI pipelines with multimodalities like text, image, audio etc. H
• ave implemented in real-world Chat bots or conversational agents at scale handling different data sources. E
• xperience in developing Image generation/translation tools using any of the latent diffusion models like stable diffusion, Instruct pix2pix. E
• xpertise in handling large scale structured and unstructured data. E
• fficiently handled large-scale generative AI datasets and outputs.
  M

L/DL Skills : H

• igh familiarity in the use of DL theory/practices in NLP applications C
• omfort level to code in Huggingface, LangChain, Chainlit, Tensorflow and/or Pytorch, Scikit-learn, Numpy and Pandas C
• omfort level to use two/more of open source NLP modules like SpaCy, TorchText, fastai.text, farm-haystack, and others
  N

LP Skills : K

• nowledge in fundamental text data processing (like use of regex, token/word analysis, spelling correction/noise reduction in text, segmenting noisy unfamiliar sentences/phrases at right places, deriving insights from clustering, etc.,) H
• ave implemented in real-world BERT/or other transformer fine-tuned models (Seq classification, NER or QA) from data preparation, model creation and inference till deployment
  P

ython Project Management Skills F

• amiliarity in the use of Docker tools, pipenv/conda/poetry env C
• omfort level in following Python project management best practices (use of setup.py, logging, pytests, relative module imports,sphinx docs,etc.,) F
• amiliarity in use of Github (clone, fetch, pull/push,raising issues and PR, etc.,)
  C

loud Skills and Computing : U

• se of GCP services like BigQuery, Cloud function, Cloud run, Cloud Build, VertexAI, G
• ood working knowledge on other open source packages to benchmark and derive summary E
• xperience in using GPU/CPU of cloud and on-prem infrastructures S
• killset to leverage cloud platform for Data Engineering, Big Data and ML needs.
  D

eployment Skills : U

• se of Dockers (experience in experimental docker features, docker-compose, etc.,) F
• amiliarity with orchestration tools such as airflow, Kubeflow E
• xperience in CI/CD, infrastructure as code tools like terraform etc. K
• ubernetes or any other containerization tool with experience in Helm, Argoworkflow, etc., A
• bility to develop APIs with compliance, ethical, secure and safe AI tools.
  U

I : G

• ood UI skills to visualize and build better applications using Gradio, Dash, Streamlit, React, Django, etc., D
• eeper understanding of javascript, css, angular, html, etc., is a plus.
  M

iscellaneous Skills : D

ata Engineering: S

• killsets to perform distributed computing (specifically parallelism and scalability in Data Processing, Modeling and Inferencing through Spark, Dask, RapidsAI or RapidscuDF) A
• bility to build python-based APIs (e.g.: use of FastAPIs/ Flask/ Django for APIs) E
• xperience in Elastic Search and Apache Solr is a plus, vector databases.

esponsibilities : D

• esign NLP/LLM/GenAI applications / products by following robust coding practices, E
• xplore SoTA models/techniques so that they can be applied for automotive industry usecases C
• onduct ML experiments to train/infer models; if need be, build models that abide by memory & latency restrictions, D
• eploy REST APIs or a minimalistic UI for NLP applications using Docker and Kubernetes tools S
• howcase NLP/LLM/GenAI applications in the best way possible to users through web frameworks (Dash, Plotly, Streamlit, etc.,) C
• onverge multibots into super apps using LLMs with multimodalities D
• evelop agentic workflow using Autogen, Agentbuilder, langgraph B