The current study focuses on characterizing the auto-ignition process from turbulent fuel injection into a high-temperature, vitiated environment using the jet-in-hot-coflow (JHC) burner configuration. High-speed (10-kHz acquisition rate) optical and laser-based diagnostics are used to identify ignition kernel formation and to determine the most probable mixture fraction and temperature conditions that directly facilitate ignition. Four operating cases are studied that present variations in coflow temperature, jet velocity, and fuel type. High-speed OH* chemiluminescence (CL) imaging is used to obtain data on the spatial position and delay time (following fuel injection) of the formation of the first auto-ignition kernels. Over the limited conditions tested, the ignition delay times and heights show a strong sensitivity to temperature and fuel type, but only a mild sensitivity to jet velocity. High-resolution, kHz-rate laser Rayleigh scattering (LRS) is used to provide simultaneous mixture fraction and temperature measurements prior to and at the onset of ignition. High signal-to-noise ratios (SNR> 200) enable reliable measurements of mixture fraction at ultra-lean conditions that seemingly promote auto-ignition. A statistical characterization of the most probable mixture fraction values leading to ignition (ξ i g) is presented and compared with calculated values of the most reactive mixture fraction, ξ M R, which has been identified previously from theory and simulations as the parameter governing auto-ignition. Results show that very lean conditions near ξ M R are preferentially encountered, but the probability density function (PDF) of ξ i g is described by a near-exponential distribution, spanning values across the flammability limits. Some limitations to the current methodologies for determining ξ i g are discussed, which are due primarily to the fact that the discretely sampled mixture fraction data is inherently asynchronous with the ignition event itself.