Datacenter Mechanical & CoolingLab-first · Mentored · Portfolio-backed

Datacenter Mechanical & Cooling Systems Internship

AI datacenters are cooling-bound — turn your Thermodynamics, Heat Transfer and RAC courses into the skill that removes 40 kW of heat from a GPU rack without melting it.

8 modules18 labs3 formatsCredit-mappable

Overview

What this internship makes you able to do.

Every Mechanical engineering student in India studies Thermodynamics, Heat Transfer, Fluid Mechanics and Refrigeration & Air-Conditioning. Almost none of them know that these four courses are exactly the skill stack the AI-datacenter boom is desperate for. A single NVIDIA GB200-class rack dissipates over 100 kW — more heat per square metre than almost any machine a fresh graduate has ever analysed — and India is building datacenter capacity from roughly 1.4 GW today toward 4 GW+ by 2030. Every one of those megawatts of compute is a megawatt of heat, and someone has to move it: through cold plates, coolant loops, air handlers, chillers and cooling towers. That someone is a mechanical engineer, and there are far too few who can do it.

This internship takes the theory you already passed exams on and makes you apply it to real datacenter problems. You start with the heat-load mathematics of actual GPU server spec sheets, size airflow with psychrometrics against the ASHRAE TC9.9 thermal envelope, and work through room-level cooling: CRAC/CRAH selection, hot/cold-aisle containment, and CFD simulation of airflow so you can see recirculation and bypass instead of guessing at them. Then you go where the industry is going — direct-to-chip liquid cooling, coolant distribution units, rear-door heat exchangers and immersion cooling for 30–50 kW/rack AI rows — and out to the plant: chillers, cooling towers, pumping schemes, free cooling for Indian climates, PUE optimisation and N+1 redundancy design. Every week produces a calculation workbook, a simulation, or a design document that a hiring manager can actually evaluate.

The internship is built for the Indian academic calendar and the AICTE/NEP internship mandate. Take it as a 4-week winter sprint, an 8-week summer internship, or a 6-month final-semester capstone that maps to your project/internship credits. Every track ends the same way: a graded capstone — a complete rack-to-room cooling design for a high-density AI rack row, defended live before a mentor panel — a portfolio of engineering calculations and CFD evidence, an RKR completion certificate, and, for the strongest interns, a referral into the mission-critical MEP and datacenter-operator hiring pipeline.

Built on your syllabus

The courses this internship extends.

You've already studied these. Here's how each one becomes a deployable skill.

Engineering ThermodynamicsMech · Mechatronics · Aero · Auto
AICTE model / Anna Univ ME3391 / JNTU & VTU Thermodynamics

The first law, control volumes and vapour-compression cycles you solved on paper become the energy balance of a live datacenter — from a 700 W GPU to a 1,000-ton chiller plant, it is the same bookkeeping, now with a PUE target attached.

Heat and Mass TransferMech · Mechatronics · Chemical
AICTE model HMT / Anna Univ ME3691 / VTU Heat Transfer

Conduction, convection and heat-exchanger effectiveness stop being derivations and become cold-plate sizing, rear-door heat-exchanger selection and the LMTD calculations behind every CDU datasheet you will read on the job.

Fluid Mechanics and MachineryMech · Civil · Chemical
AICTE model FM / Anna Univ CE3391 / VTU Fluid Mechanics

Bernoulli, head loss and pump curves turn into real work: sizing chilled-water pumps, balancing coolant loops, and reading a fan curve to fix a pressure-starved cold aisle.

Refrigeration and Air-ConditioningMech · Mechatronics
AICTE model RAC / Anna Univ & VTU professional elective

Psychrometrics and the vapour-compression cycle become CRAC/CRAH engineering — sensible-heat-ratio maths, ASHRAE envelope compliance, and knowing when an economiser can cut a Mumbai facility's compressor hours.

Choose your format

Matched to the Indian academic calendar.

Winter Internship
4 weeks
20 hrs / week · Virtual — live evening mentoring + cloud simulation lab

Credit: Fits a 2–4 week AICTE winter/vacation internship; certificate + logbook for internal credit

Best for: Pre-final year Mechanical students wanting a fast, intense first exposure to the datacenter domain

Summer Internship
8 weeks
25 hrs / week · Hybrid — live mentoring, cloud CFD/simulation lab, weekly design reviews

Credit: Maps to the standard 6–8 week AICTE summer internship required between 3rd and 4th year

Best for: The core track — 3rd-year students building a placement portfolio for mission-critical MEP and datacenter roles

Semester Capstone Internship
24 weeks
18 hrs / week · Hybrid — sustained design project with a dedicated mentor

Credit: Maps to the NEP 2020 full-semester / final-year internship-project credits (often 12–20 credits)

Best for: Final-semester students doing internship-in-lieu-of-project who want a defensible thermal-design capstone

The curriculum

8 modules. 18 labs. Week by week.

This is the full plan for the 8-week track (the winter and semester formats compress or extend the same arc). Every week ends in a deliverable your mentor reviews.

Week 1

From Thermodynamics to the AI rack: heat load

Re-anchor the first law where it now lives — inside a GPU server. Build the heat-load model of a real AI rack from component spec sheets and discover why 'cooling-bound' is the defining constraint of AI infrastructure.

You'll do
  • Decompose a real GPU server (8×H100-class) spec sheet into a component-level heat budget; roll it up to rack and row
  • Build a Python/spreadsheet heat-load calculator: IT load, lighting, people, envelope gains, safety factors
  • Compare a 6 kW enterprise rack with a 40 kW AI rack and quantify exactly where air cooling breaks down
Deliverable: Heat-load workbook for a 10-rack AI row, with every watt traced to a datasheet
Week 2

Psychrometrics & air-side fundamentals

The RAC course, weaponised. Work the psychrometric chart against the ASHRAE TC9.9 envelope and size the airflow a rack actually needs — the calculation every datacenter mechanical interview asks for.

You'll do
  • Plot supply/return states on the psychrometric chart for A1-envelope operation (18–27 °C recommended range)
  • Size airflow per rack from the sensible-heat equation across ΔT choices; see why higher ΔT is free money
  • Analyse a Chennai vs Bengaluru vs Mumbai ambient profile and its effect on cooling design and economiser hours
Deliverable: Airflow & psychrometric sizing report for the reference row, defended against ASHRAE TC9.9
Week 3

Room-level cooling: CRAC, CRAH & the chilled-water loop

The machines that move the heat: DX CRAC vs chilled-water CRAH, where each fits, and how the room connects to the plant. Map your RAC vapour-compression theory onto real unit datasheets.

You'll do
  • Trace the refrigeration cycle through a CRAC datasheet: capacity, SHR, EER at Indian ambient conditions
  • Select and size CRAH units for the reference row with N+1 redundancy; place them on a room layout
  • Model supply/return water temperatures and flow for the CRAH coil using effectiveness-NTU from your HMT course
Deliverable: CRAC/CRAH selection & sizing document with N+1 unit count and layout drawing
Week 4

Airflow management, containment & CFD

Air does not go where you assume — it goes where pressure sends it. Learn hot/cold-aisle containment, then prove your design in CFD instead of arguing about it.

You'll do
  • Design hot-aisle containment for the reference row: blanking panels, brush strips, floor-grille placement
  • Run a room-level CFD simulation; identify recirculation and bypass and quantify them as wasted kW
  • Iterate the design and show the before/after temperature field at server inlets
Deliverable: CFD report: baseline vs contained design, with inlet-temperature compliance evidence
Week 5

The plant: chillers, cooling towers & free cooling

Follow the heat all the way outdoors. Size a chilled-water plant — chillers, cooling towers, pumps and piping — and find the free-cooling hours an Indian climate actually gives you.

You'll do
  • Size chillers in kW/ton terms for the facility load; compare air-cooled vs water-cooled at Indian ambients
  • Select a cooling tower from range/approach requirements; compute water consumption and WUE
  • Design the pumping scheme (primary/secondary) with head-loss calculations from your Fluid Mechanics course, and evaluate water-side economiser hours for two Indian cities
Deliverable: One-line chilled-water plant schematic with equipment sizing and free-cooling analysis
Week 6

Liquid cooling for AI: direct-to-chip, CDUs & immersion

Above ~30 kW/rack, air alone loses. Design the technology the AI buildout runs on: cold plates, coolant distribution units, rear-door heat exchangers and immersion — the scarcest skill in the Indian market.

You'll do
  • Size a direct-to-chip loop for a 40 kW rack: cold-plate ΔT, coolant flow, CDU capacity and approach temperature
  • Compare rear-door heat exchangers, D2C (70–80% heat capture) and immersion on capacity, cost, and serviceability
  • Design the liquid/air split for a hybrid rack and the facility-water interface per OCP/ASHRAE liquid-cooling classes
Deliverable: Liquid-cooling retrofit proposal for the reference row, with CDU sizing and flow schematic
Week 7

PUE, redundancy & running the plant

Design is half the job; operating without ever going down is the other half. PUE/WUE accounting, N+1/2N redundancy, failure-mode analysis, and the monitoring that catches trouble early.

You'll do
  • Build a component-level PUE model of your facility; find and rank the three biggest efficiency levers
  • Run a failure-mode walkthrough: lose a CRAH, a pump, a chiller — compute ride-through time from thermal mass
  • Map sensor points into a DCIM/BMS monitoring plan and write the operating setpoints and alarm thresholds
Deliverable: PUE model + redundancy and failure-mode analysis with a monitoring point list
Week 8

Capstone: design, simulate, justify, defend

A fresh written brief — a high-density AI rack row — and everything you have built: heat load, airflow, plant, liquid cooling, redundancy, PUE. Design it end-to-end and defend it live, like a real design review.

You'll do
  • Produce the complete rack-to-room cooling design from the capstone brief
  • Back every major decision with a calculation, a simulation result or a standards citation
  • Present and defend the design and its failure analysis to a mentor panel
Deliverable: Capstone design package + recorded defence
Tools & tech you'll use
Psychrometric analysis · ASHRAE TC9.9 thermal envelopesAnsys Fluent / OpenFOAM (CFD fundamentals)Cadence 6SigmaDCX-class datacenter airflow simulationPython 3 · CoolProp · pandas for thermal modellingAutoCAD / Revit MEP (layout & schematics)Chiller-plant & cooling-tower selection tools (kW/ton, approach, range)Direct-to-chip / CDU / immersion cooling design references (OCP, ASHRAE liquid-cooling guidelines)DCIM & BMS concepts (Schneider EcoStruxure IT, Sunbird) · Uptime Institute Tier framework

The capstone

Cooling a High-Density AI Rack Row

You are handed a written brief: a colocation operator is deploying a 10-rack AI row at 30–50 kW per rack in an Indian metro, inside an existing facility with a chilled-water plant at 80% capacity. You must design the complete cooling solution — rack-level liquid/air split, row containment and airflow, CDU and plant-side integration, redundancy and ride-through, and a defensible PUE target — then defend it live in a design review.

Component-traceable heat-load model for the row, with safety factors stated and justified
Rack-level cooling architecture (direct-to-chip + air, RDHx or immersion) with equipment sizing calculations
Row and room airflow design validated by a CFD result showing ASHRAE-compliant server inlet temperatures
Plant-side integration: added chiller/tower/pump capacity, flow and ΔT budgets, and water consumption
N+1 redundancy analysis with computed thermal ride-through for the worst single failure
A PUE target with the component-level model that proves it is achievable, plus an operating setpoint sheet
How it's graded: Graded against a published rubric on engineering correctness, standards compliance, redundancy reasoning and the live defence. A pass earns the RKR Datacenter Mechanical certificate; a distinction earns a fast-track referral into the RKR mission-critical hiring pipeline and the certification ladder.

Measurable outcomes

Walk out able to do this — on record.

Build a component-traceable heat-load model for an AI rack row and size airflow against the ASHRAE TC9.9 envelope using psychrometrics

Select and size CRAC/CRAH units, chillers, cooling towers and pumps for a datacenter load, with N+1 redundancy

Design hot/cold-aisle containment and validate it with room-level CFD, quantifying recirculation and bypass

Design a direct-to-chip liquid-cooling loop — cold-plate ΔT, coolant flow, CDU sizing — for a 30–50 kW AI rack, and choose defensibly between D2C, rear-door and immersion

Build a component-level PUE/WUE model, identify the top efficiency levers, and compute thermal ride-through for single-failure scenarios

Produce and defend an engineer-grade cooling design document in a live review

What you keep

Your portfolio artifacts.

Heat-load & airflow calculation workbook

Component-to-row heat budgets, psychrometric sizing and airflow calculations for a real AI rack row — the engineering arithmetic interviewers actually probe.

CFD airflow analysis report

Room-level simulation evidence showing you can find and fix recirculation and bypass, with before/after inlet-temperature compliance against ASHRAE TC9.9.

Liquid-cooling retrofit design

A direct-to-chip/CDU design for a 40 kW rack with flow schematics and equipment sizing — the single scarcest skill in India's AI-datacenter buildout.

Capstone cooling design package

An engineer-grade design document for a high-density AI rack row: heat load to heat rejection, redundancy, PUE targets, and every decision justified.

RKR completion certificate

Verifiable certificate stating the graded outcome and hours — mappable to your AICTE/NEP internship credit.

Mentorship
  • Assigned mentor who is a practising datacenter mechanical / critical-facilities engineer, not a content narrator
  • Weekly live design review of your calculations, simulations and drawings
  • Async help channel with 1-business-day response on blockers
  • Interview-prep session: how to walk a panel through a cooling design like a working engineer
Evaluation & certificate

Continuous assessment on weekly deliverables (60%) plus a graded, defended capstone (40%). Every intern receives a verifiable RKR completion certificate with the graded outcome and logged hours, formatted for AICTE/NEP internship-credit submission. Distinction-grade interns receive a letter of recommendation and priority access to the RKR hiring pipeline.

Career plan

Where this internship takes you.

India's datacenter buildout is hiring mechanical engineers faster than colleges produce ones who know the domain. Cooling is the binding constraint of every AI facility, and a graduate who can carry a design from rack heat load to plant sizing — with CFD and liquid-cooling literacy — skips the generic production/site-engineer detour entirely. This internship is engineered to make you that graduate, with a graded capstone as the proof.

Roles unlocked
Graduate Engineer Trainee — Datacenter Mechanical / Critical FacilitiesHVAC Design Engineer (Mission-Critical / MEP consultancies)Datacenter Operations Engineer (Mechanical shift engineer)Thermal / CFD Analysis Engineer (datacenter & electronics cooling)
Entry band (post)
Rs 3.5–7 LPA entry, with a credible 8–15 LPA step within 2–3 years for engineers who own liquid-cooling and plant-design scope
Stipend
Merit stipend during the internship for distinction-track interns; performance-based project stipend on the semester capstone

Conversion: Distinction-grade interns are referred into RKR's hiring-partner pipeline — datacenter operators, colocation providers and mission-critical MEP consultancies staffing India's 2026–2030 buildout — and fast-tracked for the RKR certification ladder.

Rung 1 · 0-1 yr
Datacenter Mechanical GET / Facilities Engineer
Rs 3.5-6 LPA
Rung 2 · 1-3 yrs
HVAC / Datacenter Mechanical Engineer
Rs 6-12 LPA
Rung 3 · 3-6 yrs
Cooling Design Lead / Sr. Critical Facilities Engineer
Rs 12-25 LPA
Rung 4 · 6+ yrs
Mission-Critical MEP Design Manager / DC Chief Engineer
Rs 22-45 LPA
Demand signal

As of June 2026, CBRE pegs India's live datacenter capacity near 1.4 GW with 850+ MW under construction toward a projected 4 GW+ by 2030, on the back of Rs 2 lakh crore+ of committed investment. With AI racks pushing 40–130 kW each, operators consistently cite mechanical/cooling engineers as their scarcest hire, and datacenter-specific HVAC and liquid-cooling skills draw a clear premium over generic building-services roles.

8 modules. 18 labs. One credit-mappable certificate.

Build it on real gear, defend a capstone, and walk into placements with proof.