Backend Engineer, ML systems

GALAXEYE SPACE SOLUTIONS PRIVATE LIMITED; Bengaluru; Presencial; INR 900 000 - 1 300 000; Tempo integral

Resumo da oferta

GalaxEye Space Solutions Private Limited seeks an early-career backend/ML engineer to build production-grade services for multi-sensor satellite data. You’ll implement self-hosted ML workflows, offline inference, and robust data pipelines that run without internet and within defense-grade environments.

You’ll learn fast, debug hard problems, and grow into a role at the intersection of production systems, ML, and geospatial analytics.

Qualificações

• 1–2 years of backend engineering and ML engineering experience.
• Strong fundamentals: reasoning about requests and responses across systems.
• Proven debugging ability and fast learning in hard constraints.
• Evidence of rapid self-learning on challenging topics.

Responsabilidades

• Build and maintain backend services and APIs for satellite data and geospatial analytics.
• Design data flows for large imagery datasets and scalable storage.
• Deploy self-hosted ML models and production services without cloud access.
• Develop offline inference pipelines with latency/throughput considerations.
• Improve analyst tooling and ensure robust monitoring and evals.
• Debug systems where correctness is nuanced and track impact of changes.

Conhecimentos

Backend engineering

ML engineering

Debugging

Self-learning

Ferramentas

Docker

Kubernetes

GPU inference

Descrição da oferta de emprego

GALAXEYE SPACE SOLUTIONS PRIVATE LIMITED | Full time

GalaxEye is a Space-Tech startup pioneering the world's firstMulti-Sensor Earth Observation Satellite, integrating SAR (Synthetic ApertureRadar) and MSI (Multi-Spectral Imaging) on a single platform. As we movetowards building a constellation of indigenous satellites, we are alsodeveloping advanced data platforms that fuse satellite data, AI analytics, andgeospatial intelligence.

About the role

You’ll help buildthe backend and ML platforms that turn multi-sensor satellite data intogeospatial intelligence — and you’ll build them to run fully air-gapped and offline , deployed inside defense andintelligence environments with no internet access.

This is an early-career role. We're not expecting you to havedone all of this already. We're hiring for how fast you learn, how you debugwhen things are murky, and whether you have the instincts to become genuinelygood at a rare intersection: production backend, applied ML, andhard-constraint systems that have to work without the cloud crutches mostengineers lean on.

What makes this different

Most backend/ML jobslet you reach for a managed service when things get hard — a hosted model API,cloud autoscaling, pip install at deploy time. Hereyou can't. Systems run on isolated, on-prem hardware with no internet atruntime. That means:

• Models are self-hosted andrun locally — no external inference endpoints.
  
• Dependencies are mirrored andbuilds are reproducible and offline-friendly.
  
• Deployment and updates happenthrough controlled, secure processes, not push-to-cloud.
  
• Monitoring, logging, andevals all have to be self-contained.
  

If that sounds like a fun constraint rather than anannoyance, you'll fit well here.

What you’ll actually do

• Build and maintain backendservices and APIs that fuse satellite data (SAR + MSI) and servegeospatial analytics to analysts.
  
• Design and work withdatabases and data flows for large raster/imagery datasets — model thedata, write queries and pipelines that hold up at scale.
  
• Put ML models behindreliable, self-hosted services: take something that works in a notebookand make it a monitored, production service that runs offline.
  
• Work on inference and MLpipelines for imagery/geospatial analytics — batching, latency vs.throughput, GPU constraints, keeping things healthy in an air-gappeddeployment.
  
• Contribute to analyst-facingand agentic tooling — components that chain steps, call local tools, andsupport intelligence workflows (and fail gracefully when they don't).
  
• Debug systems where"correct" is fuzzy, and build the evals and monitoring that tellus whether a change actually made things better.
  

What you’ll learn here

Because this mattersas much as the work:

• How to build production MLand data systems under real constraints — offline, on-prem, security-first— a skill very few engineers ever develop.
  
• How to reason about systemsthat are probabilistic-ally correct, not just pass/fail.
  
• Applied geospatial/EO machinelearning and multi-sensor data fusion, mentored by [a lead with MLengineering experience / the team].
  
• How to stay sharp withoutmanaged services — reading source code and papers, and self-hosting whatothers just call an API for.
  

Requirements

What we’re looking for

Genuinely required:

• ~1-2 years of backend engineering / MLengineering experience.
  
• Strongfundamentals —you understand why , not just which framework method to call. You can reason aboutwhat happens between a request arriving and a response leaving.
  
• Systematicdebugging. Whensomething breaks, you form a hypothesis, reproduce it, and narrow it down— you don't just try random fixes. (This matters double when you can'tgoogle your way out live.)
  
• Evidence youlearn fast and on your own. You've picked up something hard recently and canexplain both the thing and how you learned it.
  
• You surfaceblockers early and communicate clearly when you're stuck.
  
• Comfort withconstraints and process — security discipline, careful data handling, and working withinan air-gapped environment are part of the job, not obstacles to routearound.
  
• Curiosityabout ML/AI withsome hands-on exposure — ideally you've run a model locally/self-hosted rather than only via a cloudAPI.
  

Bonus (nice to have)

Any of these are aplus — we don't expect all or even most:

• Geospatial / remote-sensingexperience: GDAL, rasterio, QGIS, working with satellite imagery, SAR, ormulti-spectral data.
  
• Self-hosting or serving MLmodels on-prem (e.g. Triton, ONNX Runtime, local LLMs, vLLM).
  
• Experience with air-gapped,on-prem, or high-security deployments; reproducible builds; offlinepackage mirroring.
  
• Familiarity withcontainerization for isolated environments (Docker/K8s), and GPU-basedinference.
  
• Exposure to observability,evals, or testing non-deterministic systems.
  
• Comfort reading a paper'smethod section or an unfamiliar library's source.