Maharjan Consulting

Services

Specialized engineering for hard technical problems

Five areas of deep expertise. Each engagement is scoped to your problem — not sold by the hour.

Scientific Computing

Design and optimize the computational infrastructure behind research — HPC clusters, workflow engines, and large-scale data processing.

The problem

Research teams often outgrow the systems they started on. Jobs queue for days, pipelines fail silently at scale, and compute costs climb without a clear owner. The bottleneck stops being science and starts being infrastructure.

Who it's for
  • University research computing groups
  • Core facilities and shared HPC environments
  • Labs processing genomic, imaging, or simulation data
  • Teams migrating between on-prem clusters and cloud
Typical deliverables
  • HPC and cluster architecture review with prioritized recommendations
  • Workflow orchestration setup (Slurm, Nextflow, Snakemake, WDL)
  • Parallelized, resumable data-processing pipelines
  • Job scheduling, resource, and cost tuning
  • Runbooks and documentation your team can maintain
Expected outcomes
  • Multi-fold reduction in end-to-end pipeline runtime
  • Lower compute spend through right-sizing and scheduling
  • Reproducible, restartable jobs that survive failures
  • Researchers unblocked from infrastructure firefighting
Typical engagement
  1. 1Profile current workloads and identify bottlenecks
  2. 2Design target architecture and migration path
  3. 3Implement, benchmark, and validate against real data
  4. 4Hand off with documentation and knowledge transfer
Slurm
Nextflow
Snakemake
WDL
Singularity
MPI
Dask

Bioinformatics

Reproducible multi-omics pipelines and scientific software — from microbiome and metagenomics to production-grade analysis tooling.

The problem

Analysis code that works once on a laptop rarely survives contact with new datasets, collaborators, or publication review. Results are hard to reproduce, pipelines are brittle, and every new project restarts from scratch.

Who it's for
  • Academic labs and translational research groups
  • Biotech R&D and computational biology teams
  • Microbiome, genomics, and multi-omics programs
  • Groups preparing pipelines for publication or regulatory review
Typical deliverables
  • Reproducible Nextflow / Galaxy pipelines with containerized tools
  • Microbiome, metagenomics, and multi-omics workflow development
  • Version-pinned, testable scientific software packages
  • Data provenance, QC, and validation reporting
  • Pipeline documentation suitable for methods sections
Expected outcomes
  • Analyses that reproduce exactly across machines and time
  • Faster turnaround on new datasets and collaborations
  • Publication- and audit-ready methods and provenance
  • Reduced dependence on a single person's local setup
Typical engagement
  1. 1Review current analysis and reproducibility gaps
  2. 2Refactor into containerized, parameterized pipelines
  3. 3Validate outputs against known results
  4. 4Document and transfer ownership to your team
Nextflow
Galaxy
Snakemake
Bioconda
Docker
Python
R

Cloud Infrastructure

Scalable, automated cloud systems on AWS and Google Cloud — with cost control, security, and infrastructure-as-code baked in.

The problem

Cloud bills grow faster than usage, environments drift from documentation, and deployments become risky manual events. Without automation, scaling means more fragility, not more capacity.

Who it's for
  • Startups scaling beyond their initial architecture
  • Research groups moving workloads to the cloud
  • Teams without a dedicated infrastructure engineer
  • Organizations facing rising or opaque cloud costs
Typical deliverables
  • Cloud architecture design and migration plans
  • Infrastructure-as-code (Terraform) and CI/CD pipelines
  • Kubernetes and container orchestration setup
  • Cost analysis with concrete optimization actions
  • Monitoring, logging, and alerting foundations
Expected outcomes
  • Meaningful reduction in monthly cloud spend
  • Repeatable, reviewable, one-command deployments
  • Elastic capacity that scales with real demand
  • Infrastructure that matches its documentation
Typical engagement
  1. 1Audit current architecture, cost, and risk
  2. 2Design target state and phased migration
  3. 3Automate with infrastructure-as-code and CI/CD
  4. 4Validate, monitor, and document the system
AWS
Google Cloud
Kubernetes
Docker
Terraform
GitHub Actions

Artificial Intelligence

Production AI systems — LLM applications, retrieval, document processing, and structured extraction that hold up outside the demo.

The problem

AI prototypes are easy; reliable AI systems are not. Hallucinations, unstructured outputs, and unpredictable costs keep promising demos from reaching production. The gap is engineering, not model access.

Who it's for
  • AI startups building on top of foundation models
  • Teams automating document and data workflows
  • Organizations with unstructured data to extract or search
  • Companies moving an LLM prototype toward production
Typical deliverables
  • LLM applications with evaluation and guardrails
  • Retrieval-augmented generation (RAG) and semantic search
  • Document processing and structured data extraction pipelines
  • Workflow automation with human-in-the-loop checkpoints
  • Cost, latency, and quality monitoring
Expected outcomes
  • Reliable, structured outputs you can build on
  • Reduced manual document and data-entry effort
  • Predictable inference costs and latency
  • A clear path from prototype to production
Typical engagement
  1. 1Define the task, success metrics, and evaluation set
  2. 2Prototype and measure against real inputs
  3. 3Harden with guardrails, retries, and observability
  4. 4Deploy and establish ongoing quality monitoring
OpenAI
Anthropic
LangChain
Vector DBs
Python
FastAPI

Software Engineering

Production software built to last — APIs, data platforms, and web applications with the testing and rigor of a senior engineering team.

The problem

Internal tools and research software often carry no tests, no documentation, and one person who understands them. As the team grows, that debt turns into a liability that slows every new feature.

Who it's for
  • Scientific software companies and research groups
  • Startups needing senior engineering without a full hire
  • Teams inheriting or hardening legacy systems
  • Organizations building data-intensive applications
Typical deliverables
  • Production APIs and services (FastAPI, Django)
  • React / Next.js web applications and dashboards
  • PostgreSQL data models and query optimization
  • Automated testing, CI, and code-quality tooling
  • Architecture reviews and technical documentation
Expected outcomes
  • Maintainable software with real test coverage
  • Faster, safer iteration for the whole team
  • Systems that survive team and staffing changes
  • Clear architecture and documentation
Typical engagement
  1. 1Assess requirements, constraints, and existing code
  2. 2Design the architecture and delivery milestones
  3. 3Build iteratively with tests and reviews
  4. 4Deploy, document, and support the handoff
Python
FastAPI
Django
React
Next.js
PostgreSQL

How we work together

A clear path from problem to solution

Every engagement follows the same transparent process — starting with a free consultation and ending with a system your team can own.

01

Free Discovery Consultation

A complimentary conversation to understand the problem before proposing anything.

02

Technical Assessment

A grounded recommendation on the right approach — not a sales pitch.

03

Custom Proposal

A transparent proposal so you know exactly what you're getting and why.

04

Implementation & Support

Collaborative delivery, with documentation and ongoing support when you need it.

Engagement & pricing

How pricing works

Every project is unique. Consulting engagements are customized based on technical requirements, project scope, timeline, and desired outcomes. Schedule a free consultation to discuss your challenges and determine the right approach.

What pricing is based on

  • Project scope and technical complexity
  • Expected outcomes and business impact
  • Timeline and urgency
  • Depth of specialized expertise required
  • Level of ongoing support

Engagements are priced by scope and outcome — not by the hour. Every engagement begins with a free initial consultation.

Invoicing is handled per the agreed proposal, typically via ACH or wire transfer, with card payment available for smaller engagements.

Project Engagement

A defined problem with a clear scope, timeline, and deliverables. Best when you know what needs to be built or solved.

Technical Assessment

A focused review of your architecture, pipelines, or costs, delivered as prioritized, actionable recommendations.

Ongoing Advisory

Retained access to senior expertise for architecture decisions, reviews, and hands-on help as your systems evolve.

Book a free consultation

Have a hard technical problem?

Every engagement begins with a free consultation. Tell me what you're working on and we'll figure out the right approach together — no obligation.