AI / ML Engineer

AI
/ ML Engineer

AI
& Intelligence Team · SecNinjaz
Technologies LLP

Position
Details

Role AI
/ ML Engineer

Team AI
& Intelligence

Location Delhi,
India (Netaji Subhash Place) — on-site with hybrid flexibility

Employment Full-time,
permanent

Experience 1
– 4 years of hands-on AI/ML delivery

Compensation As
per industry standards, commensurate with experience

About
SecNinjaz

SecNinjaz
Technologies LLP is a Delhi-headquartered AI-native cybersecurity and
technology firm serving enterprises, governments, and intelligence
customers. Our AI portfolio delivers AI-native services and sovereign
products across multiple verticals: advanced AI agents, agentic
workflow automation, agentic AI development, AI/LLM security testing,
AI red-teaming, AI SOC automation, AI integration, and applied AI for
imagery, language, and OSINT.

Founded
in 2018. Four in-house AI products anchor our work: NAINA (imagery
narrative analytics), Kaalix AI (AI-augmented VAPT), WebMine
(AI-powered OSINT across 250+ sources), and textr (post-quantum
encrypted messenger). Certifications held: ISO/IEC 42001:2023 (AI
Management System), ISO/IEC 27001:2022, ISO 9001:2015, ISO/IEC
20000-1:2018, ISO/IEC 27701:2025, and CMMI Level 3.

The
Role

This
is a hands-on engineering role. You will design, build, and
productionise ML and LLM systems that power our products and internal
automation. You will own components end to end — problem framing,
dataset work, evaluation, deployment, and monitoring — often in
environments that require on-prem, air-gapped, or otherwise sovereign
deployment. Close collaboration with security engineers, product
engineers, and the GRC team is the norm, not an exception.

SecNinjaz
separates AI-augmented from AI-native and does not overclaim. That
standard applies inside the team as well: measured, not claimed.

Key
Responsibilities

• Design
  and ship production ML, LLM, and multi-agent systems across Kaalix
  AI, WebMine, NAINA, and internal AI automation.
  

• Build
  advanced AI agents — planning, tool use, multi-agent
  orchestration, and evaluation loops — using LangGraph, CrewAI,
  AutoGen, or equivalents.
  

• Deliver
  across AI verticals: AI/LLM security testing, AI red-teaming, AI SOC
  automation, agentic workflow automation, agentic AI development, and
  AI integration.
  

• Build
  RAG and hybrid retrieval pipelines that hold up under adversarial
  and out-of-distribution inputs.
  

• Fine-tune,
  distil, or adapt open-weight models for sovereign, on-prem, or
  air-gapped deployments where API-only options are not acceptable.
  

• Design
  evaluation harnesses, red-teaming suites, and drift monitoring for
  AI systems; report performance in numbers, with error bars, not
  adjectives.
  

• Own
  the MLOps around your work: reproducible training, versioned
  artefacts, containerised inference, observability, and rollback.
  

• Contribute
  to AI governance aligned to ISO/IEC 42001 (AI Management System) and
  DPDP Act 2023.
  

• Mentor
  interns and junior AI engineers, review pull requests, and write
  internal AI documentation that outlives the person who wrote it.
  

Must
Have

• M.Tech
  in Computer Science, AI/ML, or a related discipline.
  

• 1
  – 4 years of hands-on experience delivering ML or AI systems to
  production; independent work and strong internships count if the
  outcome shipped and you can defend the design.
  

• Strong
  Python; solid working knowledge of PyTorch (preferred) or
  TensorFlow.
  

• Real
  experience building LLM applications and AI agents: prompt design,
  RAG, fine-tuning, tool use, and agentic frameworks (LangChain,
  LlamaIndex, LangGraph, CrewAI, AutoGen, or equivalents).
  

• Working
  knowledge of vector databases (FAISS, pgvector, Weaviate, or Milvus)
  and rigorous evaluation and benchmarking techniques for AI systems.
  

• MLOps
  fundamentals: Docker, model registries, CI/CD, and at least one
  orchestrator (Kubernetes, Airflow, or similar).
  

• Comfortable
  in Linux, with REST/gRPC APIs, and with distributed-systems basics.
  

• Clear
  written and spoken English; the ability to explain AI trade-offs to
  non-AI stakeholders and to write documentation that engineers
  actually read.
  

Nice
to Have

• Multi-agent
  orchestration experience — planner-worker, supervisor-subordinate,
  or graph-based agent systems in production.
  

• Applied
  AI in security — AI red-teaming, LLM security testing, AI SOC
  automation, or AI-augmented VAPT.
  

• Computer
  vision experience (object detection, tracking, multi-modal models)
  relevant to NAINA.
  

• On-prem,
  air-gapped, or edge deployment experience for AI workloads.
  

• Familiarity
  with AI safety, adversarial ML, prompt injection defence, or model
  watermarking.
  

• Model-serving
  optimisation: quantisation, GGUF/ONNX, vLLM, TGI, or Triton.
  

• Open-source
  contributions or published research in AI/ML.
  

Why
SecNinjaz

• Real
  ownership. Small,
  senior team. You decide how a system gets built, and you are
  accountable for how it performs in production.
  

• Sovereign
  work. A
  large share of what we build has to run without a public-cloud
  dependency. That constraint forces good engineering.
  

• AI
  grounded in reality. We
  separate AI-augmented from AI-native and refuse to overclaim. That
  discipline applies internally too.
  

• Audited
  process. ISO/IEC
  27001, 9001, 20000-1, 27701, 42001 and CMMI Level 3 mean the
  processes around your work are audited, not folklore.
  

• Domain
  edge. You
  will work at the seam of cybersecurity and AI — a rare combination
  outside of a few dedicated labs.