Job Description: Forward Deployed Engineer — AI/ML, GenAI, MLOps, LLMOps & AIOps
Role overview
As a Forward Deployed Engineer (FDE), you embed with Kalo's Tech Studio customers to take AI, ML and GenAI systems from first conversation to reliable production. You sit between the customer, our product team and our platform, writing real code against real data rather than slideware.
You will scope high-value use cases, build working prototypes in days, and then harden them with solid MLOps, LLMOps and AIOps practices so they stay accurate, observable and cost-efficient after you leave. What you learn in the field flows back into our product roadmap.
Key responsibilities
Customer engagement and solutioning
- Run discovery workshops with business and technical stakeholders to find AI use cases with clear ROI.
- Translate ambiguous problems into technical designs, success metrics and delivery plans.
- Act as the trusted technical advisor on-site or remote, from proof of concept to production rollout.
- Present architectures, trade-offs and results to engineers and executives alike.
AI/ML and GenAI development
- Build and fine-tune ML models (classification, forecasting, anomaly detection, NLP, recommendation) on customer data.
- Design and ship GenAI applications: RAG pipelines, AI agents, tool calling, structured extraction and copilots.
- Engineer prompts, retrieval strategies, chunking and embeddings; choose between hosted LLMs, open-weight models and fine-tuning.
- Write production-grade Python services and APIs that integrate with customer systems and data sources.
MLOps and LLMOps
- Set up CI/CD for models and prompts, including versioning, experiment tracking and model registries.
- Build evaluation harnesses for LLM apps: golden datasets, LLM-as-judge, regression tests and human review loops.
- Deploy and scale models and inference endpoints on cloud and Kubernetes, with attention to latency and GPU cost.
- Implement guardrails for safety, PII handling, prompt-injection defence and responsible AI requirements.
AIOps and production reliability
- Instrument models and LLM apps for observability: tracing, drift detection, quality,
latency and token-cost dashboards.
- Apply AI to IT operations: log and metric anomaly detection, alert correlation, root-cause analysis and automated remediation.
- Own incident response for deployed AI systems and write clear runbooks and postmortems.
Product feedback and enablement
- Turn repeated field patterns into reusable components, accelerators and reference architectures.
- Feed customer insights and feature requests to product and engineering.
- Train customer teams so they can own and extend what you built.
Must have
Candidates must meet all of these to move past screening.
- 5+ years in software, ML or data engineering, including 2+ years shipping ML or GenAI systems to production.
- Strong Python and software engineering fundamentals: APIs, testing, Git, code review.
- Built at least one production LLM application (RAG, agents or copilots) with prompt engineering and evaluation.
- Solid ML foundations: supervised and unsupervised learning, feature engineering, model evaluation.
- Hands-on MLOps: Docker, CI/CD, model deployment behind an API, monitoring and retraining.
- Production experience on at least one major cloud (AWS, Azure or GCP).
- Strong customer-facing communication; able to lead meetings and explain trade-offs to non-technical stakeholders.
- Willing to travel to customer sites up to [25–50]%.
- Bachelor's degree in Computer Science, Engineering, Mathematics or equivalent practical experience.
Skills matrix: must have vs good to have
Area
Must have
Good to have
Languages
Python, SQL
TypeScript, Go or Java
Machine learning
scikit-learn plus PyTorch or TensorFlow
XGBoost, Hugging Face Transformers, fine-tuning (LoRA / QLoRA)
GenAI and LLMs
At least one major LLM API (Claude, GPT,
Gemini); RAG; prompt engineering
LangChain, LlamaIndex, LangGraph, DSPy; MCP and tool calling; open-weight models (Llama, Mistral); multi-agent systems
Retrieval
One vector store (e.g. pgvector, Pinecone)
Weaviate, Qdrant, Milvus, OpenSearch; hybrid search and reranking
MLOps
Docker, Git, CI/CD, experiment tracking (MLflow or Weights & Biases)
Kubeflow, SageMaker, Vertex AI, Azure ML, DVC, Feast
LLMOps
LLM evaluation basics (golden datasets, LLM-as-judge, regression tests)
LangSmith, Langfuse, Arize Phoenix, Ragas, DeepEval, promptfoo, guardrail frameworks
Model serving
Deploying a model behind an API (FastAPI or similar)
vLLM, TGI, Triton, Ray Serve, KServe; quantisation and inference cost tuning
AIOps and observability
Logging and metrics basics (Prometheus / Grafana or cloud-native tools)
OpenTelemetry, Datadog, Splunk, ELK; ML for anomaly detection, alert correlation and root-cause analysis
Infrastructure and cloud
One cloud: AWS, Azure or GCP
Kubernetes, Terraform, Airflow, Spark, Kafka; AWS Bedrock, Azure OpenAI
Customer skills
Explicit communication, stakeholder presentations
Consulting, pre-sales or forward deployed experience
Good to have
These are a plus, not a requirement.
- Prior forward deployed, solutions engineering, professional services or consulting experience with enterprise clients.
- Fine-tuning (LoRA / QLoRA, instruction tuning, distillation) and inference optimisation (quantisation, batching, caching).
- Agentic or multi-agent systems running in production.
- Kubernetes and infrastructure-as-code (Terraform) in production.
- SRE or IT operations background, applying ML to logs, metrics and alerts (AIOps).
- Experience in regulated industries (finance, healthcare, telecom, public sector) and with GDPR, HIPAA or SOC 2.
- Certifications: AWS ML Specialty, Azure AI Engineer, Google Professional ML Engineer, or CKA.
- Open-source contributions, publications or talks in AI/ML.
Role details
Item
Detail
Title
Forward Deployed Engineer, AI/ML (Senior or Staff)
📌 FDE - AI/ ML Engineer (India)
🏢 KTS
📍 India