Two things: 1) The eurodollar implied futures rates need to be convexity-adjusted before they can be used as forward rates (futures rate = forward rate + convexity bias). 2) Discounting should be done using the OIS discount curve, not the LIBOR curve.
More specifically (and ignoring market conventions such as day count), let's say you're pricing a 1-year swap (6m fixed vs 3m floating) and let's assume that all the Eurodollar futures are perfectly aligned with the floating leg (i.e., there's no stub period and start & end dates are matched). Then step 1 is to compute the implied forward rates from the Eurodollar futures, which are $100 - \text{ED prices} - \text{convexity adjustments}$, where the convexity adjustments can be obtained using simple models or from dealers. Then the par swap rate is solved from $$ \frac{c}{2} \cdot d(0.5) + \frac{c}{2} \cdot d(1.0) = F_{0,0.25}\cdot 0.25 \cdot d(0.25) + F_{0.25,0.5}\cdot 0.25 \cdot d(0.5) + F_{0.5,0.75}\cdot 0.25 \cdot d(0.5) + F_{0.75,1}\cdot 0.25 \cdot d(1), $$$$ \frac{c}{2} \cdot d(0.5) + \frac{c}{2} \cdot d(1.0) = F_{0,0.25}\cdot 0.25 \cdot d(0.25) + F_{0.25,0.5}\cdot 0.25 \cdot d(0.5) + F_{0.5,0.75}\cdot 0.25 \cdot d(0.75) + F_{0.75,1}\cdot 0.25 \cdot d(1), $$
where $c$ is the par coupon rate you're solving for, $F_{t_1,t_2}$'s are the forward rates between $t_1$ and $t_2$, and $d(t)$ is the OIS discount factor from time $t$ back to the settlement date.