Doppler Radar DSP Engineer — Speed Measurement & Event Detection (Must Have Real Radar Experience)

We're building a low-cost 24 GHz Doppler radar device that measures the speed of fast-moving sports objects — speed-gun class accuracy for ball speed and ball-off-bat speed. We have a working hardware path (24 GHz I/Q Doppler, RFbeam-class module) and a commercial reference unit to benchmark against. We need a signal-processing engineer who genuinely knows radar — specifically Doppler / CW / FMCW — to build and tune the pipeline with us. This is a specialist radar role. General DSP experience (audio, comms, image processing) is not enough on its own. If your radar experience is only simulations or coursework, this isn't the right fit — we need someone who has processed real radar returns from real hardware. What you'll build: Accurate speed extraction from a 24 GHz I/Q Doppler return — FFT-based velocity estimation, peak detection, noise and clutter handling Reliable target detection across the device's full working range, using speed and distance data Robust event detection: reliably telling a genuine ball-strike apart from other motion in the beam (e.g. a bat swinging through and missing the ball), with near-zero false readings Output validated against a commercial reference unit, to match or beat its accuracy and reliability You must have: Hands-on experience with raw radar I/Q / ADC data from real modules — not just MATLAB simulations Direct experience with Doppler, CW, or FMCW radar specifically Moving-object detection, FFT, peak detection, signal classification, and tracking in high-noise environments MATLAB or Python proficiency for algorithm development Experience matching performance against a reference device How the project is structured (milestones, fixed-price): Paid trial task — a small, defined task on sample radar data so we can both confirm it's a fit before the main work begins. Milestone 1 — Speed accuracy: ball speed and straight-on ball-off-bat speed matching the reference unit within an agreed tolerance, across a set number of logged reps. Milestone 2 — Detection & whiff rejection: reliably registering valid hits/pitches while rejecting swing-and-miss events (near-zero false readings) to an agreed threshold, across a set number of test swings. The main development is an intensive on-site sprint (around October, based in Bangalore) — working together in person for fast, live iteration and testing. Accommodation and local costs covered, plus a performance bonus on hitting the accuracy and detection milestones. Open to candidates across India willing to work on-site for the sprint. To apply — you MUST do both of these. Applications that skip either will be rejected without review: Answer this directly in your proposal: Using a single Doppler radar with no camera, how would you distinguish a ball being struck and flying away from a bat swinging through the beam and missing the ball? Be specific about what features of the signal you would key on. Name the single closest radar project you have personally built or shipped — what it measured, what hardware/module you used, and your specific role in the signal processing. Please don't send a generic proposal — I'm looking for the radar detail in points 1 and 2. That's how I'll decide who to talk to.