Nervesparks
Artificial Intelligence · Machine Learning · Deep Tech
Job Description — Senior Software Engineer (AI / ML)
ROLE
Sr. Software Engineer (AI/ML)
EXPERIENCE
3 – 4 Years (Relevant)
LEVEL
L2 / L3 (based on profile)
LOCATION
On-site — Delhi
ABOUT NERVESPARKS
Nervesparks is an AI-first technology company operating at the intersection of applied machine learning research and production-grade engineering. We build intelligent systems that operate at scale — systems that reason, adapt, act, and improve over time. Our engineering culture is defined by depth, rigour, and a genuine commitment to understanding what we build, not just shipping it.
At this stage of our growth, we are looking for experienced Data and AI/ML engineers who can operate with significant autonomy — engineers who have navigated the full complexity of a machine learning system in production, who understand the gap between a working model and a reliable product, and who can lead technical decisions while continuing to contribute hands-on.
THE OPPORTUNITY
This is a senior individual contributor role for an engineer with 3–4 years of hands-on experience in applied machine learning and data-intensive systems. You will own the design and delivery of complex Data and AI/ML pipelines — including agentic AI systems, end-to-end ML pipelines, and large-scale data infrastructure — and will be expected to raise the technical bar of the team around you.
You are not here to execute a spec. You are here to define the right approach, challenge assumptions, build what works in production, and mentor the engineers around you as the team scales.
WHAT YOU WILL DO
Agentic AI Systems
› Architect, build, and maintain agentic AI systems — multi-step, tool-using, goal-directed agents capable of reasoning, planning, and taking autonomous action in complex environments.
› Design and implement agent orchestration frameworks: task decomposition, tool integration, memory management (short-term and long-term), and feedback loops.
› Work with LLM-based agents (LangChain, LangGraph, AutoGen, CrewAI, or custom orchestration) — evaluating agent reliability, hallucination risk, and output consistency.
› Define and instrument evaluation frameworks for agentic systems: task completion rates, decision quality, latency, and safety constraints.
Machine Learning & Model Development
› Lead the end-to-end development of ML models — from problem framing, data analysis, and feature engineering through to model training, evaluation, and productionisation.
› Work across ML domains as the product demands: NLP/LLMs, computer vision, time-series forecasting, recommendation systems, or structured data modelling.
› Drive hyperparameter optimisation, architecture selection, and systematic experimentation — maintaining clear documentation of what was tried and why.
› Evaluate and integrate pre-trained foundation models, fine-tune them on proprietary data, and apply techniques such as RLHF, DPO, LoRA, or QLoRA where relevant.
ML Pipelines & MLOps
› Design, build, and maintain robust ML pipelines — covering data ingestion, feature engineering, model training, evaluation, deployment, and monitoring.
› Implement MLOps best practices: experiment tracking (MLflow, Weights & Biases), model versioning, reproducible training pipelines, automated retraining, and model drift detection.
› Deploy models as production services using containerisation (Docker, Kubernetes) and cloud ML platforms (AWS SageMaker, GCP Vertex AI, Azure ML), ensuring reliability, scalability, and low-latency inference.
› Build CI/CD pipelines for ML — automating testing, validation, and deployment of model updates with appropriate rollback mechanisms.
› Monitor production models — define performance baselines, set up alerting, detect data and model drift, and manage the retraining lifecycle.
Data Infrastructure — Databricks & Snowflake
› Work with Databricks as a primary unified analytics and ML platform — building and maintaining Delta Lake tables, orchestrating workflows with Databricks Jobs and Workflows, and leveraging MLflow within the Databricks ecosystem.
› Use Databricks for large-scale distributed data processing (Apache Spark) — writing optimised PySpark and Spark SQL for data transformation, feature computation, and model training at scale.
› Integrate with Snowflake for structured data storage, transformation, and serving — writing efficient Snowflake SQL, managing schemas and warehouses, and working with Snowpark for Python-based data transformations within Snowflake.
› Design and implement feature stores — bridging offline (Databricks / Snowflake) and online (low-latency serving) feature computation for real-time ML inference.
› Collaborate with data engineering teams to ensure data quality, lineage tracking, and governance across the ML data stack.
Technical Leadership & Collaboration
› Provide technical direction on system design, architecture decisions, and tooling choices — backed by clear reasoning and trade-off analysis.
› Conduct rigorous code reviews and set standards for code quality, testing, documentation, and reproducibility within the ML codebase.
› Mentor junior engineers — providing technical guidance, reviewing their work, and helping them grow in depth and autonomy.
› Engage with product and business stakeholders to translate requirements into well-scoped ML problem statements, and communicate technical constraints and timelines with clarity.
EDUCATIONAL QUALIFICATIONS
Qualification
Notes
Required
B.Tech in Artificial Intelligence & Machine Learning, Computer Science, or a closely related engineering discipline
Strong mathematical and theoretical foundations expected. Candidates should be able to read ML research papers and translate them into implementation.
Preferred
M.Tech in Artificial Intelligence, Machine Learning, Data Science, or equivalent postgraduate qualification in a related field
Postgraduate candidates with thesis or research experience in applied ML, agentic systems, or large-scale data are especially well-suited to the depth of work in this role.
EXPERIENCE REQUIREMENTS
Required: 3–4 years of hands-on, professional experience in applied machine learning, data science, or AI engineering — with demonstrable ownership of systems that have reached production.
› At least 2 years of experience working directly with ML pipelines in a production environment — not just model development, but deployment, monitoring, and maintenance.
› Demonstrated experience with agentic AI systems, LLM-based applications, or autonomous AI pipelines — either in a product or research context.
› Hands-on experience with Databricks: Delta Lake, Spark-based data processing, MLflow on Databricks, and Databricks Workflows.
› Working proficiency with Snowflake: SQL development, Snowpark, warehouse management, and integration with upstream/downstream data systems.
› Experience with MLOps tooling and practices: experiment tracking, model versioning, automated retraining, drift monitoring, and production model management.
› Solid track record of working with cloud ML platforms — AWS SageMaker, GCP Vertex AI, or Azure ML — including managed training, endpoint deployment, and monitoring.
› Experience working in collaborative engineering environments: version-controlled codebases, code reviews, CI/CD workflows, and cross-functional team collaboration.
TECHNICAL SKILLS
Domain
Skills & Technologies
Languages
Python (advanced proficiency required). SQL (Snowflake SQL, Spark SQL). Familiarity with Scala or Java for Spark development is a plus.
Agentic AI & LLMs
LangChain, LangGraph, AutoGen, CrewAI, or equivalent orchestration frameworks. Prompt engineering, RAG (Retrieval-Augmented Generation), tool-use, memory systems, and agent evaluation. Fine-tuning (LoRA, QLoRA, RLHF, DPO). OpenAI / Anthropic / open-source model APIs.
ML Frameworks
PyTorch (strongly preferred). TensorFlow / Keras. Hugging Face Transformers, Datasets, PEFT. Scikit-learn for classical ML.
Data Platforms
Databricks (Delta Lake, PySpark, Databricks Workflows, MLflow on Databricks). Snowflake (SQL, Snowpark, Streams, Tasks). Apache Spark. Familiarity with dbt for data transformation is a plus.
ML Pipelines
End-to-end pipeline design: data ingestion → feature engineering → training → evaluation → deployment → monitoring. Apache Airflow, Prefect, or Databricks Workflows for orchestration. Feature stores (Feast, Tecton, or Databricks Feature Store).
MLOps
MLflow (tracking, registry, model serving). Weights & Biases. Model versioning and lineage. Drift detection (Evidently AI, WhyLogs, or custom monitoring). A/B testing and shadow deployment. Retraining pipelines.
Cloud & Infrastructure
AWS (SageMaker, S3, Lambda, ECR), GCP (Vertex AI, BigQuery, GCS), or Azure (Azure ML, ADLS). Docker and Kubernetes for containerised ML workloads. REST API development for model serving (FastAPI, Flask).
Software Engineering
Git (advanced). CI/CD pipelines (GitHub Actions, Jenkins, or equivalent). Unit and integration testing for ML code. Clean, modular, well-documented codebase practices.
Mathematics
Strong applied foundations: linear algebra, probability & statistics, optimisation theory. Ability to read, interpret, and implement from ML research papers.
RESEARCH, PUBLICATIONS & NOTABLE PROJECTS
While this is a role for an experienced practitioner, Nervesparks continues to value intellectual engagement beyond day-to-day delivery. The following are actively noted in our evaluation:
Area
What We Value
Published Research
Papers at recognised AI/ML venues (NeurIPS, ICML, ICLR, CVPR, ACL, EMNLP, AAAI, or similar), arXiv preprints, or workshop papers demonstrating ongoing engagement with the research community. Papers on agentic systems, LLMs, MLOps, or scalable ML are particularly relevant.
Production-Grade Projects
Publicly documented projects that demonstrate end-to-end ownership: from data pipeline design through model development, deployment, and monitoring. GitHub repositories with clean code, documentation, and reproducible results are given serious weight. We are looking for projects that show architectural thinking, not just working models.
Open-Source Contributions
Contributions to AI/ML open-source libraries — Hugging Face, LangChain, Apache Spark, MLflow, or similar — demonstrate both technical depth and professional engagement with the wider engineering community.
Technical Writing
Technical blog posts, case studies, or internal documentation that demonstrates the ability to communicate complex ML systems clearly. Candidates who can write well about what they build are particularly valued at NerveSparks.
WHAT DIFFERENTIATES A STRONG CANDIDATE
Beyond the formal requirements, the following attributes characterise the engineers who make the most meaningful impact at NerveSparks:
› Systems thinking — you see a machine learning problem as a system, not a model. You think about data quality, pipeline reliability, model degradation, and the downstream impact of your decisions.
› Depth on agentic AI — you have built agents that actually work in production: that handle failure gracefully, stay within constraints, and improve over time. You understand where agentic systems break and why.
› Production-first mindset — you are not satisfied with a model that works in a notebook. You ask: how does this behave at scale, under distribution shift, with real users, and six months from now?
› Data platform fluency — you move comfortably between Databricks and Snowflake, understand the trade-offs between them, and can design data architectures that serve both engineering and ML requirements.
› MLOps ownership — you have set up monitoring and retraining workflows, not just handed off a model to an ops team. You understand that deployment is the beginning of a model's lifecycle, not the end.
› Technical communication — you can explain complex architectural decisions in writing, defend them in discussion, and document them so the next engineer inherits context, not just code.
› Mentorship instinct — you naturally help junior engineers improve, not through lectures but through pairing, review comments, and making space for them to think independently.
GOOD TO HAVE
› Experience with multi-agent systems: agent coordination, inter-agent communication protocols, and distributed agent architectures.
› Familiarity with vector databases for RAG pipelines: Pinecone, Weaviate, Chroma, Qdrant, or pgvector.
› Experience with model compression and edge deployment: quantisation, pruning, distillation, and ONNX export.
› Exposure to real-time ML inference: streaming data (Kafka, Kinesis), low-latency model serving, and online learning systems.
› Familiarity with responsible AI and AI governance: explainability (SHAP, LIME), fairness metrics, bias auditing, and model cards.
› Experience with graph ML, causal inference, or reinforcement learning in applied settings.
› Prior experience in a startup or high-growth technology environment — comfort with ambiguity, rapid iteration, and wearing multiple hats.
HOW TO APPLY
Send your application to [[email protected]] with the subject line: Senior Software Engineer (AI/ML) — Application. Please include:
› Your CV / resume — with clear indication of the systems and platforms you have worked on and your specific contribution to each.
› Links to your GitHub, portfolio, or any production projects you can share — we read these before we read your CV.
› If you have published papers or technical writing, please include links or attach PDFs.
› A short technical note (under 300 words) describing the most complex ML system you have built end-to-end: what the problem was, what you built, what broke, and what you would do differently.
Nervesparks is an equal opportunity employer. We evaluate candidates on the quality of their thinking and the depth of their work — not on pedigree or background. All qualified applicants are encouraged to apply.
NerveSparks · nervesparks.com · [Delhi, India] · [email protected]
This document is confidential and intended solely for recruitment purposes.
Pay: ₹500,000.00 - ₹600,000.00 per year
Benefits:
- Leave encashment
- Paid sick time
Work Location: In person
