Artificial Intelligence Senior Associate

TekWissen; Chennai District; Hybrid; INR 1,200,000 - 1,800,000; Full time

Job summary

TekWissen is hiring for an Artificial Intelligence Senior Associate in Chennai. The role is hybrid (4 days in office) and requires leading AI/ML tasks with strong data governance and compliance emphasis.

You will ensure robust release practices, pair with engineers, and deliver multiple use cases with evidence packs, while mentoring teams across global time zones.

Qualifications

• Senior Associate with 3–5 years in relevant field.
• Bachelor's or Master's degree required.
• Experience in AI/ML and production-grade GenAI/agentic solutions preferred.

Responsibilities

• Apply controls on data classification and PII handling with audit trails.
• Involve cyber, legal, compliance and risk early to avoid release delays.
• Pair with client teams so deployments can run independently.
• Leave documentation, runbooks, test sets and monitoring for the receiving team.
• Handover all production items with owners within the term.

Skills

AI/ML

Python

GitHub

GCP

Data Analytics

Automation

CI/CD

ETL

Jira

Communication

Education

Bachelor's Degree

Master's Degree

Tools

GitHub Copilot

Cursor

Vertex AI

Gemini

BigQuery

Dataflow

Dataproc

Terraform

Airflow

Postgres

Job description

Overview: TekWissen is a global workforce management provider throughout India and many other countries in the world. The below client is a global company with shared ideals and a deep sense of family. From our earliest days as a pioneer of modern transportation, we have sought to make the world a better place - one that benefits lives, communities and the planet

Job Title: Artificial Intelligence Senior Associate

Location: Chennai

Work Type: Hybrid (4 Days work From Office)

Position Description:

• Apply the required controls on everything you build: data classification and PII handling, access control on grounded content, defence against prompt injection, output validation and audit trails. Produce the evidence pack for each use case - intended use, data sources, test results, known limitations, monitoring and rollback - which the named the clients owner reviews and signs.
• Bring cyber, legal, compliance and risk in early enough that review is not what holds up a release.
• Work so that nothing you build depends only on you. The paired clients engineer should be able to run it without you.
• Leave documentation, runbooks, test sets and monitoring in a state the receiving team can operate, and keep the use case record current with stage, benefit, cost, dependencies and named owner.
• In the final 30 days of the term, complete handover of everything in production and confirm each item has an owner who has run it.
• Contribute to wider use of coding assistants across the development teams through pairing, worked examples and feedback on what works.
• Build small agents for the delivery cycle where they help: test generation from requirements, remediation of scan findings, or assembling release evidence.
• Pair and review with engineers, including reviewing AI-generated code with the person who prompted it.
• Slice work so something ships monthly rather than at the end of the term.
• Work is performed on the clients equipment within approved tooling and model endpoints.
• The clients data and code may not be processed through personal accounts, devices or unapproved services.
• All code, models, prompts, test sets and documentation produced are the clients property and must be left in the clients repositories.
• Access is granted on approval, limited to the use cases in scope, and revoked on exit.
• Deliverables: one use case in production by 90 days with its test set and threshold agreed; two by six months, at least one inside a tool people already use; three or more by end of term, with two running without your involvement.

Skills Required:

• AI/ML, GitHub, Python, Google Cloud Platform, Data/Analytics, Automation, Ability to communicate and work with cross-functional teams and all levels of management , Business Analysis, Business Transformation, CI/CD, ETL, Jira

Skills Preferred:

• Google Cloud Platform - Biq Query, Data Flow, Dataproc, Data Fusion, TERRAFORM, Tekton,Cloud SQL, AIRFLOW, POSTGRES, Airflow PySpark, Python, API, GCP, Financial Reporting, Leadership, Strategic Communication 5-9 years in software or data engineering, including at least 3 years hands-on in AI/ML and at least 1 year building GenAI, LLM or agentic solutions that reached production use.
• At least one system you personally built that is in production today with real users.
• Be ready to state what it replaced, how long it took from prototype to release, which parts you wrote yourself, and how you checked it was accurate enough to release. Evidence of working quickly: something taken from idea to live use inside a few months.
• Exposure to more than one type of AI work - retrieval or knowledge systems, agent-based task automation, code comprehension or generation, or a predictive model in a live decision.
• Depth in one and working knowledge of a second is enough.
• Worked on a predictive model that reached production, with an understanding of how it was validated and monitored.
• Hands-on Python, SQL and Google Cloud (Vertex AI, Gemini, BigQuery, Cloud Run or GKE).
• Daily use of GitHub Copilot and Cursor.
• Built and tested agents: tool calling, retrieval with source citation, failure handling, and a human approval step before anything consequential happens.
• Been through a security review and a production release in a regulated setting, and produced the artefacts those required.
• Productive within the first few weeks of an engagement, working against systems you cannot change.

Experience Required:

• Senior Associate Exp: 3 to 5 years experience in relevant field

Experience Preferred:

• Financial services, insurance or another regulated industry.
• Using LLMs to make sense of long-lived systems: reading source in older languages such as COBOL, writing up what it actually does, and supporting a rewrite onto current platforms.
• Particularly useful are handling inputs larger than a context window, spotting comments and documentation that contradict the code, and generating tests that compare new behaviour against old.
• Helping a development team adopt coding assistants, and measuring what changed as a result.
• Agent-driven remediation of security or code quality findings, with a developer approving before merge.
• Building dependency or data-flow maps over undocumented systems and serving them through retrieval.
• Fairness and explainability testing on models that affect customers.
• Speech-to-text ingestion of expert sessions or recorded calls into a retrieval corpus.
• Working with subject matter experts who had very little time, and preparing well enough that short sessions were sufficient.
• Publicly visible shipped work: repositories, released tools, technical writing.
• Working across India, North America and Europe time zones, and alongside managed-service partners.

Education Required:

• Bachelors Degree, Masters Degree

Education Preferred:

• Bachelors Degree

Additional Information :

• Build assigned AI use cases from agreed requirement through to production: data access, grounding, model or agent, integration, release and handover. Expect three or more use cases over the initial term rather than one large programme.
• A clients engineer pairs with you on each one and holds the sign-off; accuracy thresholds and release approval rest with the client, and you build and produce the supporting evidence.
• Write a short proposal before building starts: the current baseline, the target, what it will cost, what has to be true before release, and what gets measured afterwards. Agree those numbers with the requesting team and the paired engineer.
• Choose the simplest design that will hold up in production - rules, a query or a small model where those will do.
• Build the test set and scoring method for each use case, keep them in version control next to the code, and run them before every release.
• Report accuracy, latency and cost honestly, including where the system fails.
• A use case that is not ready should be reported as not ready. After release, watch for drift, cost and unexpected use, and fix what the monitoring shows. Build ingestion, retrieval, agent scaffolding and deployment steps so the next use case starts from working parts rather than a blank repository, and use the shared test harness rather than a separate arrangement per project.
• Deliver output inside the tools people already use rather than as a separate interface, including systems owned by other teams or vendors.
• Implement the human checkpoint agreed for each use case: what runs unattended, what needs sign-off, and what goes straight to a person.
• Work in Cursor and GitHub Copilot from week one, on Google Cloud, with Vertex AI and Gemini alongside Claude for code generation and for reading unfamiliar codebases.
• Raise problems early. If a use case will not clear its accuracy bar, or a dependency is not going to arrive, we need to hear it while there is still time to act.

TekWissen Group is an equal opportunity employer supporting workforce diversity.